You are on page 1of 31

Ann. Rev. Biophys. Biophys. Chern. 1987. 16: 319-49 Copyright «) 1987 by Annual Reviews Inc.

All rights reserved



ABSORPTION, SCATTERING, AND IMAGING OF BIOMOLECULAR STRUCTURES WITH POLARIZED LIGHT!

Ignacio Tinoco, Jr. and William Mickols

Chemistry Department and Laboratory of Chemical Biodynamics, University of California, Berkeley, California 94720

Marcos F. Maestre

Donner Laboratory, Division of Medical Physics, University of California, Berkeley, California 94720

Carlos Bustamante

Chemistry Department, University of New Mexico, Albuquerque, New Mexico 87131

CONTENTS

PERSPECTIVES AND OVERVIEW 320

INTERACTION OF POLARIZED LIGHT WITH MATTER 321

Stokes Vectors 321

Mueller Matrix 322

ABSORPTION 324

Circular Dichroism of Large Objects 324

Fluorescence Detected Circular Dichroism 325

SCATTERING 333

Circular Intensity Differential Scattering .. 333

DIFFERENTIAL IMAGING 339

Theory 339

Experiment 343

1 Research was supported in part by NIH grants GM 10840 and RR 01613 to IT, AI 08247 to MFM, and GM 32543 to CB, and by DOE grant FG03-86ER60406 to IT. CB was supported by a Searle Scholarship, a Sloan Fellowship, and the Center for High Technology Materials, University of New Mexico.

319

0883-9182/87/0610-0319$02.00

320 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

PERSPECTIVES AND OVERVIEW

To obtain the maximum amount of information from a sample using light as the probe, we must shine polarized light on the sample and analyze the state of polarization of the transmitted or scattered light. In general, biomolecular spectroscopists do not use incident polarized light and do not analyze the polarization of the light after it has interacted with the sample. Here we point out the new information obtainable in principle with polarized light, and we discuss in more detail some specific applications.

Probably the main application of polarized light in biostructural studies has been in the measurement of optical rotatory dispersion (ORO) and circular dichroism (CD). In ORO the rotation of the plane of polarization is measured as a function of wavelength. In CD the differential extinction of right and left circularly polarized light is measured. Optical rotation has the advantage that it can be measured at any wavelength; it does not require an absorption band. However, the general method of choice is CD, because its simpler spectrum can be related more easily to molecular structure. Applications to structures such as proteins, nucleic acids, membranes, viruses, and chromatin have been reviewed (36,37,51,82,86). It eventually became apparent that contributions to circular dichroism could come from differential scattering of circularly polarized light; see Reference 15 for a quantitative discussion. The measured circular dichroism has two contributions, circular differential absorption and circular differential scattering:

eL -eR = (aL -aR)+(sL -SR)·

An easy test for differential scattering contributions is to move the sample relative to the detector. Because the scattering is angle dependent, moving the sample changes the fraction of the differentially scattered light captured by the detector. This causes a change in CD with position; pure circular differential absorption should give a signal independent of the geometry of sample, light beam, and detector. The experimental separation of (aL -aR) and (SL -SR) can be done by various methods (84). In the best method one surrounds the chiral, scattering object by a fluorophore, which captures all the light not absorbed by the sample. The circular differential fluorescence of this reporter molecule is a direct measure of (aL - aR), the circular differential absorption of the chiral scatterer of interest (68). This is one application of fluorescence detected circular dichroism; more direct applications of this method are discussed in this review.

Although circular intensity differential scattering was originally considered a nuisance and an artifact that distorted the "true" CD, we now

STRUCTURE AND POLARIZED LIGHT 321

recognize that it provides valid structural information, just as circular differential absorption does. In particular, the angular dependence of the circular intensity differential scattering (CrDS) offers unique information about chiral objects the size of the wavelength of light.

A new field described here is the experimental production and theoretical interpretation of differential polarization images. The usual image in a microscope is a measure of the transmitted light intensity. We discuss images that are directly proportional to circular dichroism (eL - eR), linear dichroism (ell-e_j_), and other polarization-sensitive parameters.

The most general characterization of the linear interaction of light with matter is obtained when the incoming and outgoing light is represented by Stokes vectors and the interaction is represented by a 16-component Mueller matrix. Several books and reviews discuss this formulation from a general point of view (3, 7, 72, 75, 92). The many components of the Mueller matrix imply that potentially useful new techniques will be developed to better characterize biomolecular structures with polarized light. Extension of the wavelength region through the vacuum ultraviolet to the soft X-ray region with a synchrotron source (1) should be very valuable. Undoubtedly, the next five years will be as productive in development of new methods using polarized light as the last five years have been.

INTERACTION OF POLARIZED LIGHT WITH MATTER

Stokes Vectors

Four parameters are necessary and sufficient to characterize a monochromatic light beam; these are the four components of the Stokes vector (65). This is true for linear interaction of light with matter. Interactions involving the square of the light intensity, which can become important with very high intensity lasers, are not considered here. The components of the Stokes vector characterize the intensity and polarization of light; they are related to the electric vector of a plane light wave (92, p. 41). For light propagating along unit vector k, the (transverse) electric field is

where Re means the real part of the expression in brackets. Ex and E; are complex, oscillating functions that specify the components ofE along unit

vectors i and j. .

The Stokes parameters are:

322 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

J = E,E;+ EvE; Q = EyE~-ExE; U = EyE~+E\E~

v = i(EvE;-E,E"';),

with the asterisk denoting the complex conjugate. For fully polarized light J2 = Q2+ U2+ V2, for partially polarized light J2 > Q2+ U2+ V2, and for unpolarized light, Q = U = V = 0.

It is convenient to use normalized Stokes vectors; from the definitions of the components it is easy to represent different states of polarization of light. Some examples are:

Arbitrary

No polarization

(J, Q, u. V) (1,0,0,0) (I, 1,0,0)

(I, - 1,0,0) (I,O,±I,O) (1,0,0, ± I).

Linear polarization along y

Linear polarization along x Linear polarization ± 45°

Right ( + ) or left ( - ) circular polarization

When light interacts with anything, its intensity and state of polarization will change. The interaction is represented by a 4 x 4 matrix, the Mueller matrix, which multiplies the incident Stokes vector to produce the resultant Stokes vector.

Mueller Matrix

The intensity and state of polarization of light after it has interacted with a sample is given by

[J] [MIl

Q M21

U M31

V M41

MI2 MI3

M22 M23

M32 M33

M42 M43

Here the 16-term matrix M is the Mueller matrix. Generally all 16 components of the matrix are independent and can provide information about the sample. However, it is rare for all 16 components to be measured (5). If unpolarized light is incident and only the intensity of the transmitted light is measured, then MIl is the only component obtained. The value

STRUCTURE AND POLARIZED LIGHT 323

of M II characterizes the extinction, A, of the sample by the relation Mil = 10-A.

It is instructive to consider the independent and nonzero terms in the Mueller matrix for various types of samples to learn how much more information can be obtained by using incident polarized light and by analyzing the state of polarization of the outgoing light. For a nonchiral solution the transmitted light at zero angle is characterized by three independent components:

[~l [T

o 0

Mn 0

o M22 o 0

JJ [f:i

Therefore, I = MilIa; Q = M22QO; U = M22UO; and V = M44VO' In addition to the extinction coefficient (related to Mil), one can measure the depolarization of incident linearly polarized light (Mn) and the depolarization of incident circularly polarized light (M 44)' For a chiral solution two additional parameters are needed to characterize the transmitted light at zero angle:

o

o

o

The circular dichroism is characterized by MI4 and the optical rotation by M 23' If we measure the scattered light from a chiral solution at any angle (except 0 and 180°), 10 independent components can be obtained:

[~i=[P:

U MI3

V MI4

For a nonrandom (oriented) chiral sample all 16 components of the Mueller matrix are needed to characterize the scattered light completely. Experimentally, this is a difficult task (5, 72, 78, 79), but with increasing understanding of the significance of each component (11, 12, 30-32, 56- 58) there is increasing impetus to measure all 16 values.

In this review we concentrate on the first row of the Mueller matrix; these are the easiest parameters to measure. A linear or circular polarizer

324 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

is used in front of the sample, but only the intensity of the transmitted or scattered beam is measured. We thus emphasize circular dichroism (M14 at 0°), circular intensity differential scattering (M14 at any angle), and linear dichroism (M,Z' MI3 at 0°). The theory and practice of imaging based on differential absorption and scattering of polarized light are also discussed.

ABSORPTION

Circular Dichroism of Large Objects

Circular dichroism ofbiomacromolecules has been reviewed often in recent years (36, 37, 51, 82, 86). Therefore we describe only topics not covered earlier. A theoretical understanding of the CD for chromatin, membrane complexes, condensed DNA, or even nuclei of cells and whole cells has been lacking. Of course, in principle one could first obtain the wave functions for these large objects, then use standard equations to calculate their CD (81, 88). This is obviously not practical. Realistic models and methods are necessary. One likely approach is the classical coupled oscillator method of DeVoe (18, 19). Here an object is treated as a collection of point polarizable groups. Incident light induces oscillating dipoles in each group. The induced dipoles are considered to interact as static dipoles with a dependence on r: 3. This method is very appropriate for calculating the CD of proteins of nucleic acids, but it is not sufficient for treating aggregates that are large in three dimensions compared to the wavelength of light. Keller & Bustamante (38, 39) have shown that for large objects the full electrodynamic interaction among the induced dipoles must be considered. The concept that the optical properties (absorption and circular dichroism) of an object can be calculated from coupled oscillators is unchanged; the improvement is in the interaction term that couples them. The interaction tensor is (38):

eikr ieikr eikr 1

rex, x') = (3ii-1) 4nk2r3 -(3ii-1) 4nkrz + (1 -if) 4nr - 3kz <5\r).

Here r = x-x', r = [r], i = rjr, and k = 2n/A, with A equal to the wavelength of light. The first term depends on r: 3; it decreases rapidly with distance so that the exponential can be set equal to 1. The first term is essentially the static dipole-dipole interaction term. The second term is an r: Z -dependent term called the intermediate contribution. The third term is the radiation field interaction, which depends on r: I. The last term simply ensures that the self interaction is zero. For small objects (kr « 1) or for objects that are large (compared to the wavelength of light) in only

STRUCTURE AND POLARIZED LIGHT 325

one or two dimensions, the static dipole-dipole term dominates. For large objects (kr = I) all three terms are generally important. Keller & Bustamante (38) found the eigenmodes of the coupled linear equations for the induced dipoles. They showed that eigenmodes with right or left helical symmetry will interact preferentially with right or left circularly polarized light. Kim et al (43) have done model calculations that give CD spectra similar to those of psi-type CD measured for condensed DNA.

Electrodynamic coupling among point polarizabilities is the same procedure that several groups are using to calculate all the scattering components of the Mueller matrix (10, 56, 76, 95).

Tobias (87) has proposed a phenomenological, macroscopic theory of the CD of nucleosomes. He related the CD to the linear dichroism and linear birefringence of oriented DNA molecules on the nucleosome. This type of model could be extended to other arrangements of linearly anisotropic structures.

Fluorescence Detected Circular Dichroism

Fluorescence detected circular dichroism (FDCD) is the use of fluorescence to monitor the preferential absorption of circularly polarized light; the method was developed by Turner et al (90, 91). The emitted fluorescence is measured instead of the transmitted light as in CD. The fluorescence is then recorded in the standard way by a CD instrument so that the apparent ellipticity in degrees, e, read on the chart can be related to the FDCD. The fluorescence intensity is related to the properties of the fluorophore in a cell of length d by (85):

1.

where 10 is the incident intensity, ¢ is the fluorescence efficiency of the fluorophore (quantum yield), eF is the molar absorption coefficient of the fluorophore, CF is the concentration of fluorophore, and A is the total absorption of the cell, including absorption by the fluorophore and other nonfluorescent species. When the ratio (FL - FR)/(FL + FR) is calculated from the above expression, the measured apparent ellipticity is:

2.

R is the correction term for the overall CD and absorbance of the sample and is given by:

3.

326 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

AL and AR are the total absorbances for the solutions for left and right circularly polarized light and are computed as follows: AL = A + 8/65.96 and AR = A-8/65.96. Equation 2 is composed of two terms: a concentration-independent term, i1SF/SF = gF, which is the Kuhn dissymmetry factor (44), and the R term, which is a function of the total absorbance and CD of the sample. For the typical FDCD measurement the values of gF and R are on the order of 10-3 and the equation reduces to:

4.

where R can also be approximated as

i1A 2.303i1A IO-A

R = --- - ----~

2A 2(l-IO-A)

5.

and i1sF is the molar CD of fluorophore. Thus the Kuhn dissymmetry factor for the fluorophore, gF = i1SF/SF, can be obtained by extrapolating the measured ellipticity to zero concentration. Alternatively, R can be calculated from the CD and the total absorbance of the solution obtained in a separate measurement.

FDCD is useful in cases where biological molecules, such as proteins, nucleoproteins, or nucleic acids, have many chiral groups but only one or two fluorophores. The CD of the fluorophore can then be separated from the rest of the CD of the molecule. If there is more than one fluorescing species in solution, the measured Kuhn dissymmetry factor is the quantum yield weighted sum of the contribution of each of the species:

i1sF L cPiCii1Si

----

SF LcPiC;S;

6.

These species can be stable molecules or different conformations of a flexible molecule that are in rapid equilibrium. The measured CD and absorbance of the same system gives:

It is clear that for a rigid molecule or a molecule with only one fluorescing conformer, the CD and the CD extracted from the FDCD signal should be the same. Examples are d-IO-camphorsulfonic acid and tryptophan (90). However, for molecules that have several conformers, some that fluoresce and others that do not, the CD and FDCD are very different.

STRUCTURE AND POLARIZED LIGHT 327

For example, dinuc1eoside phosphates containing fluorescent bases (69) have at least two states in solution. The state with the bases stacked has the characteristic CD of the dimer, but the fluorescence is quenched, and therefore the FDCD measures only the un stacked conformations. It is possible to orient macromolecules by electric fields and by hydrodynamic shear (17, 80) and to measure their CD along different directions. For a molecule whose rotation is slow compared to its fluorescence lifetime, the FDCD depends on the average CD (de) and on the CD along the direction of the fluorescence emission transition vector (de33) (83). Photoselection causes the FDCD to measure the CD of the oriented molecules. If a linear polarizer is used in front of the detector, the presence of photoselection can be detected, and both de and de33 can be obtained. For a detector perpendicular to the incident light,

4deF/eF - 2de33,F/eF gF = -_ .. _--.-

3+e33,~)eF '

7.

where de33,F and e33,F are the circular dichroism and absorptivity along the direction of the fluorescent emission transition moment. If the detector is placed at an angle of 54.75° (the magic angle) to the incident light, the average CD is measured. With a perpendicular detector that has an analyzer in front, the average CD and the CD along the transition moment direction can also be obtained.

WHAT DOES FDCD MEASURE? FDCD measures the difference in absorbance of right versus left circularly polarized light of the ground state of fluorescent molecules. If there is energy transfer, the CD of molecules that transfer energy to the fluorophore is also measured. Since FDCD is an emission measurement, there are some conditions that have to be satisfied for the measurement to be valid. The first and most important condition is that the quantum yield must be independent of the state of polarization of the exciting light. A second condition is that instrumental parameters must be adjusted so that the circular polarization of the illuminating light is very pure, to avoid linear dichroism artifacts (35, 49, 50).

ARTIFACTS Spectra of molecules that exhibit isotropic fluorescence because of rapid tumbling have shown good agreement with theoretical predictions (89, 90). However, for systems that give partially polarized fluorescence because of slow rotation, FDCD spectra do not correlate well with theory. Nonchiral fluorophores such as fluorescein or rhodamine dyes give large signals when in viscous solutions such as glycerol, but no signal in aqueous solutions. These molecules are not circularly dichroic, and they should not have a signal whether they are oriented or not. Lobenstine (47)

328 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

did a series of experiments to determine the source of the artifacts and found that the most important factor was the orientation of the Pockels cell. Rotation about the direction of propagation of the light beam indicated that the artifact was probably due to residual linear polarization in the circularly polarized light beam. Small changes in the angle or magnitude of this linearly polarized component can generate large apparent FDCD signals.

Hug & Surbeck (35) have presented an instrumental fix for the problems of residual linear dichroism in experiments of Raman optical activity. The fluorescence emission is measured in a plane perpendicular to the direction of the exciting beam by two photomultiplier tubes, which are in turn perpendicular to each other. The resultant signals are electronically summed, and the artifacts due to residual linear polarization in the excitation beam are removed. The incorporation of the two photomultipliers in the FDCD machine effectively resolves the artifact problem, and the remaining difficulties are a function of the materials measured (47).

SENSITIVITY OF TECHNIQUES AND INSTRUMENTATION Fluorescence detection methods are very sensitive because they enable us to measure very small signals at a particular wavelength different from the exciting wavelength. FDCD techniques provide a way to measure the circular differential absorbance of chromophores (the ground state) of only those molecules that fluoresce; the other nonfluorescing absorbers affect the signal through concentration-dependent terms that can be removed by variation of concentration. In principle, fluorescence signals of the same order as the stray light could be detected by the best instrumentation; that is, signal-tonoise ratios of about one could be measured. Lamos & Turner (45) have suggested the use of pulsed excitation methods with time-gated detection methods to overcome this limitation. Fluorescence would only be measured after the exciting light was off. Lamos & Turner and their collaborators (45, 46) have built the most sensitive FDCD machine extant, which has been shown capable of detecting FDCD signals produced by ethidium molecules at a concentration of 5 x 10-6 M when bound to 5 x 10-5 M poly(dG-dC). In the presence of nonbound ethidium, which has no CD, Lamos et al (46) could measure FDCD signals with only 10% of the total ethidium bound to the polymer. In one of their published spectra, signals obtained from 10.2 ,uM ethidium interacting with 22.1 ,uM poly(dG-dC) exhibited an r value of 0.0083, i.e. less than one ethidium molecule per 100 base-pairs of polymer.

APPLICA TIONS OF FDCD In spite of the sensitivity and specificity of the technique, FDCD has been used sparsely, probably owing to the lack of instrumentation. The technique was first applied to the study of the anti-

STRUCTURE AND POLARIZED LIGHT 329

codon loop of the yeast tRNAPhe (91), which has the fluorescent Y base (63). The results showed that FDCD was specific for conformational changes near the anticodon loop. The FDCD measurement suggested a transition in the anticodon loop near 20°C, followed by a broad transition from 30 to 60°C and finally a sharp melting of the entire RNA. The removal of Mg2+ from the RNA decreased the FDCD signal by a factor of two; this was interpreted as a decrease in conformational rigidity in the anticodon loop.

Soon after, Reich & Tinoco (69) reported FDCD measurements for three fluorescent dinucleoside phosphates containing I,N6-ethenoadenosine (eApeA, eApC, and eApU), together with their CD and absorbance measurements. This study answered some of the questions concerning the stacking properties of the dinucleoside phosphates. Since FDCD detects only the fluorescent species, and the stacked states of the three eA dimers are quenched, only the unstacked species are measured. Thus, if the molecules exist in a two-state system, there will be only one fluorescent species, and the dissymmetry factors, which are independent of concentration, should not exhibit any changes with respect to temperature. If, on the other hand, more than one state exists, then the dissymmetry factor is given by Equation 6, where ¢i now refers to the quantum yield of species i and the summation is over all fluorescent states of the dimer in solution.

Reich & Tinoco (69) thus showed that, within experimental error, the dissymmetry factor of eApC does not change with changing temperature; indeed it is a two-state system. The FDCD of eApU shows that it has a stacked state in equilibrium with a fluorescent state that is not fully unstacked. The results for eApeA are more complex; the interpretation is that this dimer does not have two-state behavior, but rather has a series of discrete steps. The more complicated behavior is also confirmed by the values obtained from similar CD analysis of the temperature melting data.

Proteins The first FDCD studies on proteins were reported by Lobenstine et al (48); the proteins adrenocorticotropic hormone, glucagon, human serum albumin, and monellin all contained a single fluorescent tryptophan. Values obtained for the Kuhn dissymmetry factor, gF, ranged from - 1 x 10-3 for human serum albumin to 1.65 x 10-3 for monellin, compared to 4 x 10-4 for free tryptophan. The magnitude of gF is a measure of the chirality of the tryptophan plus the chirality of any groups transferring energy to the tryptophan. For the proteins studied, the energy transfer was predominantly from tyrosine. Half the excitation would have been transferred within a distance of 10 A. This is a first approximation for localization of conformational changes detected by FDCD when energy transfer is present. When most of the fluorescence arises from tryptophan

330 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

absorption, the region probed by FDCD is even smaller. One advantage of FDCD that these results point out is that the gF spectra have large values in wavelength regions where both tryptophan and tyrosine have low absorbance. This is because gF is not the value of LisF but is the dissymmetry, LisF/SF; FDCD is therefore more sensitive to weak transitions than conventional CD spectra.

Photoselected fluorescence circular dichroism Lobenstine & Turner (49, 50) presented the first results on the detection of a particular polarization of the fluorescence of an optically active f1uorophore. According to Tinoco et al (83), when a detector is perpendicular to the exciting light, and a linear polarizer is placed in front of it, the measured FDCD for rigid molecules is:

[8R(1 +COS2 ¢)+(8R33/3)(2-3. cos ' ¢) J

SF = -14.32 2 -2R] ,

D(4-cos ¢)-(Dd3)(2-3 cos? ¢)

8.

where ¢ is the angle that the detected polarization makes with the axis perpendicular to the excitation beam, Rand D are the average rotational and dipole strengths, and R33 and D33 are the rotational and dipole strengths along the direction of the emission transition moment. R] here is equivalent to R in Equation 5.

Equation 8 predicts that at the polarizer angle ¢ = 35.25° the FDCD is a simple function of R, D, and the values of A and LiA obtained from the CD and total absorbance. Lobenstine & Turner (49) measured the CD and FDCD of d-lO-camphorsulfonic acid in glycerol and of morphine in glycerol and water and obtained very close fits for the measured 0F and the computed values from the CD and absorbance measurements.

In a second publication, Lobenstine & Turner (50) addressed a very important assumption which is the basis of the FDCD measurement, i.e. that the quantum yield of the fluorescence is independent of the polarization of the exciting light. Others had claimed that there was significant difference in the quantum yield for right and left circularly polarized exciting light for D- and L-tryptophan. Lobenstine & Turner, using the two-detector instrumentation that corrects for residual linear dichroism components, showed that there was complete agreement between the FDCD and the computed values from the CD and absorbance spectra for both molecules.

Studies of binding of ethidium to nucleic acids in vivo and in vitro Most nucleic acids do not fluoresce. It is possible therefore to measure the FDCD of a fluorescing compound that interacts with the nucleic acids and whose

STRUCTURE AND POLARIZED LIGHT 331

optical behavior shows CD because of the induced chirality of the binding conformation. Lamos & Turner and their collaborators (45, 46) have used this property to investigate the binding of ethidium to Escherichia coli DNA in vitro and in vivo. FDCD bands were found at 325 nm and at 385 nm. These bands changed magnitude as the binding ratio of ethidium to DNA changed. The largest negative peak at 385 nm corresponds to an ethidium transition known to have a small e but a large Kuhn dissymmetry, Se]». The positive band at 325 nm arises from the ethidium transitions centered at 335 and 300 nm. The predicted FDCD spectrum computed from the CD and absorption spectra by use of Equation 2 showed that the measured spectrum band at 385 nm was larger; this disparity was interpreted as heterogeneity in the binding sites in DNA. As the ethidium concentration was lowered, the 385 nm band increased both in vivo and in vitro.

Lamos et al (46) measured the FDCD of ethidium bound to poly(dGdC) and poly(dG-m5dC) under B- and Z-form conditions. The FDCD spectra were similar for both conditions. This implies that when ethidium binds to polymers in Z-form conditions it clusters into regions that it has induced into a right-handed form.

Differential scattering corrections One of the most useful applications of FDCD is the correction for differential scattering components of the CD signal of materials that are comparable in size to the wavelength. The structures must also have chiral superstructure. This differential scattering is measured by standard instruments as if it were differential absorption and can lead to major distortions in the spectra. The presence of such scattering can be indicated by an observed CD signal in wavelength regions where the sample does not absorb light.

Experimental methods for correction of differential scattering or total scattering depend on increasing the solid angle of detection to collect more of the scattered light. This has been accomplished by increasing the size of the detector window, bringing the cuvette closer, or using integrating devices such as the Fluorscat cuvette (20, 21). This method has been used successfully in many applications (21, 22, 34, 54).

Unfortunately the light detector cannot be placed between the sample and the incident beam in any of the above experimental methods. Thus no correction is possible for differential back scattering by the sample. However, with the proper application of FDCD, a fluorescent substance that is not optically active can be used as a CD detector with a solid angle of detection of 4n steradians. In Equation 5, AA is the total differential absorbance of the sample and A is the total absorbance. Thus although only emission from the fluorophore is measured, the signal is a function

332 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

of the optical activity of all the components of the sample. This can be understood if one considers that the relative amount of right and left circularly polarized light reaching the fluorophore at any point in the solution is mediated by the differential absorbance of the medium in front of the fluorescer.

Consider the case where the sample to be measured consists of a mixture of a non-optically active fluorescent substance and an optically active differential scatterer. In this case i1SF/SF is zero, and Equation 4 can be rearranged to:

2A(lOA-l)0F

i1A = 28.65(lOA-2.303A-l)"

9.

Therefore the CD of a differential scatterer can be measured through the use of a non-optically active fluorophore. The scattering particles are in effect surrounded on all sides by fluorescent detectors, which intercept the scattered light in addition to the incident radiation. These detectors yield a signal that is a function of all the scattered and transmitted light.

This method for correction of differential scattering has been applied successfully in studies of optically active particles. Reich et al (68) and Maestre & Reich (54) applied the method to the study of nucleic acids in ethanolic buffers in which the natural and synthetic polymers form psitype particles. They used z-naphthylamine as the fluorescent reporter molecule because it did not interact with the nucleic acids at the concentrations used (OD305 ~ 0.3). The fluorescence lifetime of z-naphthylamine is long enough for photoselection directional effects to be averaged out (83). Differential scattering components were measured in regions of the scattering envelope not previously accessible to measurement, i.e. backwards scattering was measured along the incident light beam. This technique, properly applied, will give the maximum correction available for CD measurement of differential scatterers. When Maestre & Reich (54) applied this method to the study of polymer films with mechanical twists, they were able to measure the back-scattering reflections of films of DNA and found that these films have similar properties to those of cholesteric liquid crystals or twisted nematic liquid crystals (54).

Lamos & Turner (45) used a negatively charged nonchiral dye, coumarin 175, to measure the scattering-corrected CD of samples of whole cells of E. coli strain DM800 in the regions above 300-350 nm for the determination of the value of R in Equation 2. This technique worked even in the presence of ethidium intercalated in the cell DNA.

Hall & Maestre (28) used the FDCD scattering correction to test for the possibility of differential scattering contributions in the structures

STRUCTURE AND POLARIZED LIGHT 333

formed in ethanolic buffers by the synthetic polymer poly(dGC)· poly(dGC). They found no differential scattering contributions for these aggregates; the signal measured was a true reflection of the polymer CD at high ethanol concentrations.

AVAILABLE INSTRUMENTATION Turner (89), Lobenstine (47), and Lamos & Turner (45) have published excellent descriptions of the constructions of the different versions of the PDCD machines. Recently a new CD instrument produced by Jasco, the Jasco 600, has appeared on the market with an PDCD attachment. Thus a commercial instrument can be purchased that provides the ability to measure the PDCD signal of compounds in which there are no photoselection artifacts, i.e. compounds with long fluorescence lifetimes compared to the rotational relaxation of the molecule. It is probable that with minor modifications the Jasco 600 could become a very sensitive instrument that would also correct for photoselection artifacts for most fluorescent molecules.

ADVANTAGES OF FDCD With the increased sensitivity of the PDCD instrumentation, circular dichroism of very small magnitudes can be measured. Since the magnitudes of the PDCD bands depend on gF = /).f.F/f.F, the sensitivity is particularly enhanced for transitions with small extinction coefficients but large rotatory strengths.

FDCD methods are also the only ones that correct completely for differential scattering contributions to the CD signals of structures that are large with respect to the wavelength of light. These techniques also give a good picture of the spatial distribution of the differential scattering components of the particle.

SCATTERING

Circular Intensity Differential Scattering

Within the last decade there has been a great deal of emphasis on applying traditional optical activity methods (ORD, CD) to the study of complex systems such as particulate aggregates of biological macromolecules, suspensions of whole or fractured membranes, and nucleohistone complexes. A common feature of all these systems is the presence of anomalous CD signals outside the absorption bands and extending toward the red wavelengths. Here we review the main developments in this field during the five years since our last review (82).

THEORY Circular intensity differential scattering (CrDS) is the preferential scattering of one of the circular polarizations of the light by chiral

334 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

molecules. Since the total scattering cross section of the molecule varies as a function of the scattering angle, it has been convenient to define a normalized CIDS ratio,

10.

where h(8) and h(8) are the intensities scattered at an angle (8) for incident left and right circularly polarized light, respectively.

Atkins & Barron (2) have given general expressions for the differential scattering oflight for both Rayleigh and Raman processes from a quantum mechanical point of view. In their treatment the wavelength of light was assumed to be large compared to the dimensions of the molecules, and no absorptive phenomena were taken into account. Later, Barron & Buckingham (4) presented both classical and quantum mechanical analyses for the elastic and inelastic differential scattering of circularly polarized light. Harris & McClain (30,31) have presented a theory for the polarization of light scattered by polymers. They constructed the M ueller matrix (or Perrin matrix) for the scattering of light according to Stokes formalism (65) and related this phenomenological treatment to molecular parameters. The wavelength was assumed to be large compared to the size of the monomers, but not necessarily large relative to the polymer. This treatment is general; it does not explicitly consider any particular geometry.

Other attempts to fit the Mie scattering theory with ad hoc terms to account for anomalies in the CD spectra of particulate suspensions have had quite varied degrees of success (6, 23, 25, 26, 34, 70). Bohren (6) solved the scattering of an optically active isotropic sphere within the frame of Mie theory, using phenomenological coefficients to describe the sphere's differential responses to right and left circularly polarized light.

Bustamante et al (11-13) derived a first Born approximation classical theory that applies to oriented helices. They later generalized this treatment for arbitrary geometries and random orientation of the scattering species (14). The main results of this work are:

1. CIDS results from the interference effects of wavelets generated at different points in the chiral scatterer. These wavelets maintain a welldefined phase and polarization relationship to each other. As the dimensions of the scatterer decrease relative to the wavelength of light, the efficiency of this interference effect decreases. Thus, CIDS tends to zero with a decrease in the molecular dimensions faster than regular scattering. The interference phenomenon that gives rise to CIDS for large systems is greatest when dimensions governing the chirality of the scatterer are similar to the wavelength of the incident light.

STRUCTURE AND POLARIZED LIGHT 335

2. A sufficient condition for the occurrence of eIOS from a chiral molecule is that the molecule's polarizable groups must possess anisotropic polarizability tensors. Conversely, structures composed solely of isotropic scattering groups cannot give rise to eIOS signals in the first Born approximation, even though the groups may be arranged in a chiral fashion.

3. The angular dependence of eIOS is more sensitive to the chiral parameters of the scatterer than regular unpolarized scattering, but substantially less sensitive to the overall size of the scatterer.

4. Because eros is an interference effect, in the first Born approximation the eros in the forward direction vanishes.

5. If the incident radiation falls inside the absorption band of the chiral scatterer, the eros shows anomalous scattering behavior characterized by a breakdown in the symmetry of the eros patterns.

These characteristics are generally also valid for randomly oriented chiral scatterers (14).

Higher order Born approximations have also been used in the derivation of theoretical expressions for polarized light scattering in general (56, 58) and for erDS in particular (10, 40, 95). In the first Born approximation, the effective perturbation field at each point-polarizable group in the scatterer is approximated by the incident electromagnetic field. In higher Born approximations, the internal field must be calculated from the incident field by taking into account the interactions among the dipoles induced throughout the scatterer. Two main interactions are typically considered (19): (a) short-range static dipole coupling and (b) long-range scattering coupling, which corresponds to internal multiple scattering effects inside the particle ..

Higher Born approximations are necessary in the description of optically dense systems or systems with group polarizabilities large enough for the groups to have a significant interaction. In these treatments (10, 40, 64) several new effects appear:

1. The eIDS in the forward direction does not vanish, because the interaction among the polarizable groups introduces phase retardation.

2. The eros of a chiral arrangement of point-polarizable groups that possess spherically symmetric polarizability tensors is not zero. The asymmetry in the polarizability required for eIOS is provided in this case by the coupling among groups.

3. The eIOS ratio of periodic structures depends only on the characteristics of one unit cell of the array and is independent of the lattice repeat. Strictly speaking, this property holds only for oriented structures.

336 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

4. The presence of CIDS in the forward direction is controlled by the multiple scattering interaction. Short-range dipole coupling cannot give rise to forward CIDS, because such interaction does not produce retardation effects.

McClain & Ghoul (56), Keller et al (40), and Zeitz et al (95) have made CIDS calculations using the higher Born approximations.

Keller et al (42) have derived a quantum mechanical treatment ofCIDS within the frame of quantum electrodynamics; with the appropriate approximation the results reduce to those derived using classical electrodynamics. This theory makes no approximations regarding the size of the scatterers relative to the wavelength of the light.

Contributions of CIDS to the measured CD signals The expressions of CIDS for randomly oriented scatterers have been used to calculate the contribution of CIDS signals to the circular dichroism measured inside and outside of the absorption bands (9, IS). For small CD values the signal measured in the presence of circular differential scattering is:

II.

where IR and h are the intensities transmitted for right and left circularly polarized light, c is the concentration of the chromophore, I is the pathlength of the cuvette, (h(O) and (TR(O) are the forward scattering cross sections for left and right circularly polarized light, and r is the distance from the sample to the detector. The first term in Equation II is the usual one considered in CD studies; however, in the presence of chiral scatterers this term has contributions from the absorption and scattering coefficients:

12.

The differential scattering coefficient SL -SR is related to the integral over all angles of the angle-dependent scattering cross sections, (TL(e), (TR(e). From Equation 11 we see that the CIDS contribution to the measured CD signal is provided by a concentration-dependent term proportional to the integrated differential scattering cross sections for right and left circularly polarized light, and by a term independent of concentration related to the forward differential scattering cross sections. The second term can be minimized simply by placing the photodetector far from the cell sample (large r) and by decreasing the acceptance angle of the detector (small (TL(O) , (TR(O». The scattering contribution to the CD of the first term in

STRUCTURE AND POLARIZED LIGHT 337

Equation 11 can be eliminated only by means of the fluorscat technique (21) or by FDCD corrections (68).

Similarly, the measurement of CIDS can be complicated by the substantial contributions of CD signals when the wavelength of light chosen in the CIDS experiment is close to an absorption band. To measure the CIDS signal as a function of the scattering angle the intensities of right and left circularly polarized light arriving at the scattering volume should be identical. In practice, preferential attenuation of the beam by absorption and/or scattering of one of the two incident polarizations in the cuvette will modify the true CIDS signal according to (15):

[(1 +cos 8)2]

CIDSmeas = - 2.303(SL - sR)cl 2(1 +cos if) + CIDStrue'

The true CIDS signal can easily be obtained by varying the concentration of the chiral scatterers and extrapolating to zero concentration. What is the contribution of CIDS to the optical rotatory dispersion of a sample? Bustamante (8) has recently investigated this question. Since circular dichroism and ORD are related through the Kronig-Kramers transforms, the contribution of CIDS to the apparent circular dichroism signals outside the absorption bands of the chromophores must have an associated ORD signal. Using the Kronig-Kramers relations and the optical theorem, Bustamante showed that ORD arises from the circular preferential removal of the coherent part of the incident light as it travels through the scattering medium. Through this work it has been possible to relate all manifestations of optical activity (CD, CIDS, and ORD) by means of the Kronig-Kramers relations.

CIDS of periodic and hierarchical structures Calculations on periodic structures that have identical repetitive chiral units but a different number of units show that the angular dependence of the CIDS ratio is less sensitive to the overall size of the scatterer than regular scattering (given by the denominator of Equation 10). The angular dependence of the numerator in Equation 10 [h(8)-IR(8)] is even less sensitive to scatterer size over a wide range of dimensions. Instead, this dependence appears to be controlled almost exclusively by the dimensions and parameters of the repetitive chiral unit in the structure. This property gives rise to a unique behavior of the difference [h(8) - IR(8)]; Patterson et al (64) found that values of[h(8) - IR( 8)] for chiral structures with several hierarchical levels of chirality correspond to the sum of the contributions of the different levels instead of to the product of these contributions, as is observed in crystallography. These authors have called this effect the superposition principle, and have pointed out that this property of the difference

338 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

[h(e) - fRee)] could enormously simplify the analysis of different structural levels in a macromolecular system.

EXPERIMENTS As mentioned above, the existence of CIDS was first recognized through the circular dichroism signals detected outside the absorption bands in large chiral assemblies. Maestre et al (52), using the 442 nm line from a He-Cd laser and the electronics of a modified Cary 60, measured CIDS as a function of the scattering angle for a suspension of randomly oriented sperm nuclei. The polar plots of the measured CIDS pattern show a series of scattering lobes of alternating sign as predicted by the theory (12, 14). The maximum CIDS signal was about 2%. Experimental studies on suspensions of microorganisms have also appeared in the literature (27, 71). In some of these studies CIDS was measured at right angles from the direction of the incident beam, as a function of the wavelength of light from 300-700 nm. Various suspensions of viruses and live bacteria have been studied in this manner (71). However, some concern exists about whether or not the signals measured contain mixing oflinear birefringence terms, since no precautions to avoid these effects were reported.

Wells et al (93) have recently analyzed the possible sources of artifacts in the measurement of CIDS. These authors have established the requirements of the quality of the incident circular polarizations to measure the CIDS signals unequivocally. An analysis of the effects of spurious linear components in the incident circular polarizations shows that the true CIDS signal is related to the measured signal by:

CIDSmeas = [1-1(oc+ + oc_)] CIDStrue + 1[oc+ - oc_],

where z., and «: are the percent ellipticities of the incident circular polarizations of the light. The expression shows that imperfect circular polarizations can give rise to artifacts of two different natures: One is multiplicative and is proportional to the mean ellipticity of the incident circular polarizations, (1)(oc+ +oc_). The second is additive and is proportional to the mean differential ellipticity between the two opposite circular polarizations, (1)(oc+ - oc_). It is not hard to reduce the multiplicative errors by controlling the mean ellipticity to values of a few percent. The additive error is harder to eliminate and can easily give values comparable to the magnitude of the true CIDS values. Synchronous demodulation methods must be used to eliminate this source of artifacts. Finally, a protocol for alignment and polarization quality control is proposed. These authors have substantially refined the earlier sperm cell measurements.

Liquid crystals There is a great interest in the optical properties of liquid crystal systems, for often these systems manifest optical behavior quite

STRUCTURE AND POLARIZED LIGHT 339

similar to that observed in large molecular aggregates. In particular, cholesteric liquid crystals are known to form long-range chiral structures that produce unique optical activity. Bustamante et al (16) have measured the cros of several cholesteric-nematic mixtures of CB-15 and ZLI-1612 as a function of the scattering angle and for different ratios of pitch to wavelength. They used samples in planar orientation. The authors found that the CIOS ratio ranged from a few percent to close to 100% when the wavelength of the light matched the pitch of the cholesteric liquid crystal. A comparison of the experimental results with a right-handed twisted ladder model using the second Born approximation showed good agreement. The experimental values of CIDS in the forward and backward directions at pitch to wavelength ratios close to unity are the largest values of cros so far measured at these angles. Hall et al (29) have carried out CIOS studies of cholesteric liquid crystals using the planar and focal conic orientations. Their results clearly show that chiral scatterers of opposite handedness but similar pitches have identical CIOS patterns, but with opposite sign as predicted by the theory.

DIFFERENTIAL IMAGING

Maestre & Katz (53) developed an instrument to measure the circular dichroism of microscopic systems. Maestre et al (55) then used the CO microscope to study the CD and CIOS signals of individual Chinese hamster ovary cells, synchronized at various stages of their life cycle. Their results have shown that the condensation of chromatin through the different stages of an individual cell can be followed by CD and CIOS signals. The measurement can be carried a step further by placing a lens between the object and a detector. When the images obtained using two orthogonal polarizations are subtracted, the resulting image is a mapping of the polarization-dependent optical properties of the object (59-62). Furthermore, domains that have different optical properties and different molecular structures can be distinguished and spatially resolved. This mechanism permits generation of polarization-dependent contrast between any given domain and its surrounding background. Linear dichroism images have been used to study the polymerization of hemoglobin in individual sickle cells (60).

Theory

Keller et al (41) have derived a theory of differential polarization imaging applied to the case of circular polarizations. Only chiral objects can give non vanishing differential polarization images. The sample is described as a collection of point-polarizable groups in first Born approximation

340 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

scattering theory. The lens is assumed to be thin, and the light intensity arriving at the detector is obtained using classical vector diffraction theory.

The image equations have been obtained in transmission geometry (bright-field imaging) and in scattering geometry (dark-field imaging). Differential absorption (CD) and scattering (CIDS) of the circular polarizations contribute to both bright-field and dark-field images, but their relative importance can be modified by choosing wavelengths inside or outside the absorption bands of the specimen. Figure 1 displays two large helices of opposite handedness and a small right-handed helix. Figure 2 shows the differential images calculated by subtracting the images formed with right and left circularly polarized light; they are compared with the usual intensity images. As these circular differential images are differences, they have both magnitude and sign. The images can be used to determine the handedness of the helices even when the image is not well resolved, as shown for the small helix.

The theory has recently been generalized for treatment of arbitrary states of the incident polarizations and arbitrary polarization components of the light transmitted and/or scattered by the object to form the images

Figure 1 Three helices. The two large helices are equal in size but have opposite handedness; the helix on the left is left-handed. The small helix is right-handed and is one third the size of the large helices (from 15).

STRUCTURE AND POLARIZED LIGHT 341

o r(}©> @\:0 0

cA\© ©'0 0

o r(}© ©~ 0

o r(}© @~

a

b

Figure 2 Calculated images of the helices in Figure 1: (a) the usual intensity image. (b) The circular differential image obtained by subtracting the image obtained with right circularly polarized light from that with left circularly polarized light. Positive values are indicated by solid lines; negative values are dotted lines. Note that the large helices have different sign images and that the sense of the small helix can be deduced even though its image is not well resolved (from 15).

342 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

(M.-H. Kim, L. Ulibarri, D. Keller, C. Bustamante, unpublished). This treatment utilizes the Mueller calculus in which the intensity and polarization of the light is described by the four scalar parameters that constitute the Stokes vector of light (/,Q,U, V). The interaction with the object (characterized by the Mueller matrix) modifies the polarization of the transmitted or scattered light; the result is a new Stokes vector. The 16 elements of the Mueller matrix contain all the structural information that can be obtained from the interaction of an object with light of a given wavelength and polarization. A combination of incident polarizations and polarization measurements of the light scattered or transmitted by an object can be used to obtain its Mueller matrix, as follows (4a):

[~-[~

M21 = --_

2

[h-[~ MI2=-i-·

n-:): _([h_I')

M_h_h v v

22 - 2

[I; - n - (l~ - [~)

M42 =

2

I" _[U M _:,:4545

13 - 2

[~-[I M14= 2

M03 = [:45-I'+45-(lh_45-J'-45)

" 2

[~- [~ - m - J'()

M24=---2

[+45_[-45_([+45_1 45)

M _ +45 +45 -45 45

33 - 2

r: [~-(ll-[:)

M44 =..

2

The subscripts of the intensities, I, denote the nature of the incident polarization: u = unpolarized, h = horizontal, v = vertical, r = right circular, and I = left circular. Superscripts denote the polarization component of the light that is allowed to pass through the analyzer after its interaction with the object; superscript u means no analyzer is used after the object. Thus [~ is the intensity of the light that passes through a horizontal analyzer when the object is illuminated with unpolarized light.

STRUCTURE AND POLARIZED LIGHT 343

As before, if we put a lens behind the object an image can be generated.

If at each point on this image the intensities are processed according to the above equations, the 16 Mueller matrix images of the object can be recorded and displayed in false color. These images contain novel information about the structural organization of the object, and a microscope can be designed to access this information. For example, the M 12 image is a mapping of the linear dichroism of the object when the wavelength is close to the absorption band of the object and proportional to the linear birefringence and linear differential scattering, for wavelengths outside the absorption band. Similarly, MI4 is a map of the CO and cros of the object.

Experiment

Using a microscope to image the various Mueller matrix elements allows us to obtain the maximum amount of information on a microscopic sample. The light used to image the sample does not have to contain the transmitted light, and it does not have to be of the same wavelength. Therefore, a Mueller matrix can be written for dark-field, bright-field, fluorescence, or phosphorescence imaging.

In the past, the Mueller matrix elements of large samples have been measured with rotating polarizers and analyzers; the physical motion of the optical elements limits the sensitivity of this method (33). Thompson et al (78, 79) have developed a multiple frequency modulator system that has the advantage of higher sensitivity and simultaneous measurement of all the Mueller matrix elements. Either of these approaches would work on microscopic samples.

We have chosen to construct an instrument that measures only the top row of the Mueller matrix. These elements do not require an analyzer behind the sample. We thus measure absorption (which is related to Mil), the amplitude oflinear dichroism and the angle oflargest linearly polarized absorption (related to MI2 and MI3), and circular dichroism (related to MI4)' Mickols et al (61) described the experimental separation of these parameters.

We have built three microscopes to measure polarization-sensitive differential images; they are described in Figure 3. The light source is a lamp with an IR filter to limit heating. An interference filter selects the desired wavelength. The light is polarized, and its polarization is modulated by either a Pockels cell or a photoelastic modulator.

One microscope uses a linear photo-diode array as the detector (61); it is scanned across the image plane to obtain an image. The output of the detector is converted to 12-bit intensity data at 2 MHz. The magnitudes

344 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

Eyepiece 0

~Photomultiplier Tube

'-:-1

/)

}'--l-.;o\_ b;..- - t - - - - -

Pinhole 2 Removable Second Modulator and

Analyzer

Transfer Lens

-121 Removable Standards

I---l

:~[JJ :

I 1 1--1

lUI

I 1 L __ _j

Pockels Cell

Polarizer Filter Pinhole I

Lamp

Figure 3 A diagram of the microscope used to provide differential polarization images. Moving a mirror allows the image to be seen by eye or directs it onto a photo-diode array or photomultiplier tube.

of [I, [2, and (11-[2)/(11+[2) are measured for the two polarizations (produced by a Pockels cell) at each image point. The necessary signal-tonoise ratio is obtained by repeating the measurement 256 times at each position of the diode array. The linear array has 1024 photo sites. The slowest step is the data transfer; the overall rate of image formation is approximately 30 msec per image point. This method has demonstrated a sensitivity for (II - [2)/(11 + [2) of one part in 10,000; it has been used for linear dichroism imaging of sickle cells (59, 60).

Another microscope uses an image dissector camera as the imaging device. A third microscope is much slower, but it has a greater wavelength range and a higher signal sensitivity because it uses a photomultiplier tube as the detector. The microscope uses a commercial x,y scanning stage with a resolution of 0.2 j.lm. This type of microscopy uses two pinholes. The first pinhole is at the focal point of the lamp. A second pinhole in front of

STRUCTURE AND POLARIZED LIGHT 345

the photomultiplier and in the image plane further limits the sampling area. The photomultiplier output is converted to voltage, and two lock-in amplifiers measure the signal at the polarization modulation frequency and at twice the modulation frequency. If the amplitude of the polarization modulation gives one quarter wave retardation, then the lock-in at the modulation frequency gives primarily circular dichroism, while the lockin at twice the modulation is primarily linear dichroism. Owing to defects in the optical train as well as in the phase modulation, mixing occurs between the circular and linear dichroism signals. This mixing can be measured using standards, and the linear and circular dichroism can be separated (61). This microscope design allows us to add more modulators and an analyzer to measure any element of the Mueller matrix.

Optical scanning microscopy has several advantages over microscopy using illumination over the whole field. Optical scanning gives higher photometric accuracy, since light from different parts of the observation field that normally mix owing to the optical transfer function and scattered light due to reflection off the lens surfaces is nearly eliminated (24, 94). Also, when both pinholes are small enough the resolution approaches the theoretical limit (74) for an optical microscope. However, full-field illumination and an array detector provide a great increase in speed of data acquisition, although the sensitivity of a diode array is generally less than that of a photomultiplier. Future advances in array technology will probably give the scanning method the spectral range and sensitivity that it needs to match the present advantages of a photomultiplier tube. However, non uniformities in sensitivity of the elements in a diode array will still present difficulties.

We have used the array detector microscope for circular dichroism imaging as well as linear dichroism imaging. The sample is imaged with different amplitudes for the voltage modulation on the Pockels cell. The value of (I, - 12)/(/, + 12) for each voltage is acquired before the array is moved. This provides one image that is mainly circular dichroism (l,2 are L,R) and another image that is mainly linear dichroism (1,2 are 11,.1). Because the images are in register they can be manipulated to separate the two effects. The main problem is to remove linear dichroism contributions to the circular dichroism. The effect of circular dichroism on linear dichroism is usually negligible.

LINEAR DICHROISM IMAGING OF SICKLE CELLS The red blood cell is one of the simplest intact cell systems that we can study. The cell membrane surrounds a cytoplasm that, for our optical measurement, is primarily hemoglobin. Both linear dichroism and circular dichroism occur in red

Figure 4 A set of images of deoxygenated red blood cells from a patient with sickle cell anemia (W. E. Mickols, unpublished data). The images were taken at 415 nm, which is the center of the hemoglobin Sorel band. The left-hand image is the absorption image, which shows the distribution of hemoglobin within the cells. The right-hand image is the linear dichroism image of the same cells, which shows the intracellular distribution of aligned hemoglobin. The left-hand image is coded for absorbance of 0.0 to 0.7, whereas the right-handed image is coded for a difference ill absorbance of -0.02 to +0.02.

2

o

.8

STRUCTURE AND POLARIZED LIGHT 347

blood cells (59, 60). We have used linear dichroism to image the amount and the alignment of polymerized hemoglobin within deoxygenated red blood cells from patients with sickle cell anemia (59, 60). In Figure 4 we compare the linear dichroism image (right), with the absorption image (left), taken simultaneously. The latter shows the distribution of hemoglobin, and the former the distribution of aligned hemoglobin. Color coding is required to fully show the usefulness of the images. Novel intracellular morphologies of aligned hemoglobin were found. We found several distinct classes of cells, two of which are shown in Figure 4. The rightmost cell is a classic sickle cell; it has most of the hemoglobin polymer aligned in the same direction in a single domain. The next three cells in a clockwise direction all have a central constriction in the linear dichroism image, which is not seen in the absorption image. We interpret this type of image as evidence for a single nucleation site for intracellular polymerization of hemoglobin. The next cell is only partly in the picture; it is seen in absorption, but its linear dichroism is zero. The topmost cell has less linear dichroism than the first four cells, but it also shows a central constriction. We have seen other types of hemoglobin polymerization patterns in the cells. Some cells show many small domains of hemoglobin oriented in different directions. There are cells that are nearly as round as normal red blood cells, but the linear dichroism shows that the hemoglobin is polymerized concentrically around the periphery of the cell.

By integrating the linear dichroism and absorption over the cell area, we can measure the amount of hemoglobin aligned in each cell. We are studying the effect on the amount of aligned hemoglobin of changes in intracellular concentration of hemoglobin, rate of deoxygenation, and time after deoxygenation. We can thus compare the results of in vitro experiments with those of in vivo experiments. We hope to understand what physical process controls the amount and alignment of hemoglobin intracellularly.

We have also imaged chromosomes of Drosophila melanogaster. The chromophores of interest in the chromosome are in the ultraviolet, so imaging with them is more difficult. The banding patterns are near the resolution of the wavelength oflight; thus diffraction is an important effect. The chromosome is also well known to be birefringent (73). To separate these effects we have used dyes that, when bound to condensed DNA, mimic the circular dichroism and circular differential scattering of the DNA (66, 67). The same types of circular dichroism and circular differential scattering are seen with mitotic Chinese hamster ovary cells (55) as well as viruses (82) and chromatin (77). The dyes allow us to image in the visible spectrum, where the contributions due to non-nucleic acid chromophores are minimized.

Literature Cited

348 TINOCO, MICKOLS, MAESTRE & BUSTAMANTE

1. Allen, F. S., Bustamante, C., eds. 1985.

Applications of Circularly Polarized Radiation Using Synchrotron and Ordinary Sources. New York: Plenum. 193 pp.

2. Atkins, P. W., Barron, L. D. 1969. Mol.

Phys. 16: 453

3. Barron, L. D. 1982. Molecular Light Scattering and Optical Activity. Cambridge, UK: Cambridge Univ. Press. 408 pp.

4. Barron, L. D., Buckingham, A. 0.1971.

Mol. Phys. 20: 111

4a. Bickel, W. S. 1985. See Ref. I, p. 69

5. Bickel, W. S., Davidson, J. F., Huffman, D. R., Kilkson, R. 1976. Proc. Natl. Acad. Sci. USA 73: 486

6. Bohren, C. F. 1975. PhD thesis. Univ.

Arizona, Tucson

7. Bohren, C. F., Huffman, P. R. 1983.

Absorption and Scattering of Light by Small Particles. New York: Wiley. 530 pp.

8. Bustamante, C. 1984. J. Opt. Soc. Am.

AI: 1114

9. Bustamante, C., Maestre, M. F., Keller, D. 1985. Biopolymers 24: 1595

10. Bustamante, C., Maestre, M. F., Keller, D., Tinoco, I. Jr. 1984. J. Chem. Phys. 80:4817

11. Bustamante, C., Maestre,M. F., Tinoco, I. Jr. 1980. J. Chem. Phys. 73: 4273

12. Bustamante, c., Maestre, M. F., Tinoco, I. Jr. 1980. J. Chem. Phys. 73: 6046

13. Bustamante, C., Tinoco, I. Jr., Maestre, M. F. 1981. J. Chem. Phys. 74: 4839

14. Bustamante, c., Tinoco, I. Jr., Maestre, M. F. 1982. J. Chem. Phys. 76: 3440

15. Bustamante, c., Tinoco, I. Jr., Maestre, M. F. 1983. Proc. Natl. Acad. Sci. USA 80: 3568

16. Bustamante, c., Wells, K. S., Keller, D., Samori, B., Maestre, M. F., Tinoco, I. Jr. 1984. Mol. Cryst. Liq. Cryst. 111: 79

17. Chung, S. Y., Holzwarth, G. 1975. J.

Mol. Bioi. 92: 449

18. DeVoe, H. 1964. J. Chem. Phys. 41: 393

19. DeVoe, H. 1965. J. Chem. Phys. 43: 3199

20. Dorman, B. P., Hearst, J., Maestre, M.

F. 1973. Methods Enzymol. 270: 767

21. Dorman, B. P., Maestre, M. F. 1973.

Proc. Natl. Acad. Sci. USA 70: 255

22. Girod, T. C., Johnson, W. C. Jr., Huntington, S. K., Maestre, M. F. 1973. Biochemistry 12: 5092

23. Gitter-Amir, A., Rosenheck, K., Schneider, A. S. 1976. Biochemistry 15: 3131

24. Goldstein, D. J. 1970. J. Microsc, 92: I

25. Gordon, D. J. 1972. Biochemistry 11: 413

26. Gordon, D. J., Holzwarth, G. 1971.

Proc. Natl. Acad. Sci. USA 68: 2365

27. Greg, C. T., McGregor, D. M., Grace, W. K., Salzman, G. C. 1985. Proc. 4th Int. Symp. Rapid Methods Automation Microbiol. Berlin: Springer-Verlag

28. Hall, K., Maestre, M. F. 1985. Biopolymers 23: 2127

29. Hall, K., Wells, K. S., Keller, D., Samori, B., Maestre, M. F., et al. 1985. In Applications of Circularly Polarized Radiation Using Synchrotron and Ordinary Sources, ed. F. S. Allen, C. Bustamante, pp. 77--91. New York: Plenum

30. Harris, R. A., McClain, W. M. 1977. J.

Chem. Phys. 67: 265

31. Harris, R. A., McClain, W. M. 1977. J.

Chem. Phys. 67: 269

32. Harris, R. A., McClain, W. M. 1985. J.

Chem. Phys. 82: 658

33. Holland, A. c., Gayne, G. 1970. Appl.

Opt. 9: 1113

34. Holzwarth, G., Gordon, D. J., McGinnes, T. E., Dorman, B. P., Maestre, M. F. 1974. Biochemistry 13: 126

35. Hug, W., Surbeck, H. 1979. Chem. Phys.

Lett. 60: 186

36. Johnson, W. C. Jr. 1978. Ann. Rev. Phys.

Chem. 29: 93

37. Johnson, W. C. Jr. 1985. Methods Biochem. Anal. 31: 163

38. Keller, D., Bustamante, C. 1986. J.

Chem. Phys. 84: 2961

39. Keller, D., Bustamante, C. 1986. J.

Chem. Phys. 84: 2972

40. Keller, D., Bustamante, c., Maestre, M.

F., Tinoco, I. Jr. 1985. Biopolymers 24: 783

41. Keller, D., Bustamante, c., Maestre, M.

F., Tinoco, I. Jr. 1985. Proc. Natl. Acad. Sci. USA 82: 401

42. Keller, D., Bustamante, c., Tinoco, I.

Jr. 1984. J. Chem. Phys. 81: 1643

43. Kim, M., Ulibarri, L., Keller, D., Maestre, M. F., Bustamante, C. 1986. J. Chem. Phys. 84: 2981

44. Kuhn, W. 1958. Ann. Rev. Phys. Chem. 9: 417

45. Lamos, M. L., Turner, D. H. 1985. Biochemistry 24: 2819

46. Lamos, M. L., Walker, G. T., Krugh, T.

R., Turner, D. H. 1986. Biochemistry 25: 687

47. Lobenstine, E. W. 1981. PhD thesis.

Univ. Rochester, New York

48. Lobenstine, E. W., Schaefer, W. C., Turner, D. H. 1981. J. Am. Chem. Soc. 103: 4936

49. Lobenstine, E. W., Turner, D. H. 1979.

J. Am. Chem. Soc. 101: 2205

STRUCTURE AND POLARIZED LIGHT 349

50. Lobenstine, E. W., Turner, D. H. 1980.

J. Am. Chem. Soc. 102: 7786

51. Maestre, M. F. 1984. In Optical Techniques in Biological Research, ed. D. L. Rousseau, pp. 292-341. New York:

Academic. 430 pp.

52. Maestre, M. F., Bustamante, c., Hayes, T. L., Subirana, J. A., Tinoco, I. Jr. 1982. Nature 298: 773

53. Maestre, M. F., Katz, J. 1982. Biopolymers 21: 1899

54. Maestre, M. F., Reich, C. 1980. Biochemistry 19: 5214

55. Maestre, M. F., Salzman, G. c., Tobey, R. A., Bustamante, C. 1985. Biochemistry 24: 5152

56. McClain, W. M., Ghoul, W. A. 1986. J.

Chem. Phys. 84: 6609

57. McClain, W. M., Harris, R. A. 1983. J.

Chem. Phys. 79: 3689

58. McClain, W. M., Schuarte, J. A., Harris, R. A. 1984. J. Chem. Phys. 80: 606

59. Mickols, W. E., Bustamante, C., Maestre, M. F., Tinoco, I. Jr., Embury, S. H. 1985. Bio-Technol. 3: 711

60. Mickols, W. E., Maestre, M. F., Tinoco, I. Jr., Embury, S. H. 1985. Proc. Natl. Acad. Sci. USA 82: 6527

61. Mickols, W. E., Tinoco, I. Jr., Katz, J.

E., Maestre, M. F., Bustamante, C. 1985. Rev. Sci. Instrum. 56: 2228

62. Mickols, W. E., Tinoco, I. Jr., Maestre, M. F., Embury, S. H. 1986. Biophys. J. 49: 29a

63. Nakanishi, K., Furutachi, N., Funamizu, M., Grunberger, D., Weinstein, I. B. 1970. J. Am. Chem. Soc. 92: 7617

64. Patterson, C. S., Singham, S. B., Salzman, G. c., Bustamante, C. 1986. J. Chem. Phys. 84: 1916

65. Perrin, F. 1942. J. Chem. Phys. 10: 415

66. Phillips, C. L., Mickols, W. E., Maestre, M. F., Tinoco, I. Jr. 1986. Biophys. J. 49: 1299

67. Phillips, C. L., Mickols, W. E., Maestre, M. F., Tinoco, I. Jr. 1986. Biochemistry 25: In press

68. Reich, C., Maestre, M. F., Edmonson, S., Gray, D. M. 1980. Biochemistry 19: 5208

69. Reich, c., Tinoco, I. Jr. 1980. Biopolymers 19: 833

70. Rosenheck, K., Schneider, A. S. 1973.

Proc. Natl. A cad. Sci. USA 70: 3458

71. Salzman, G. c., Griffith, J. K., Gregg, C. T. 1982. Appl. Environ. Microbiol. 44: 1081

72. Schellman, J., Jensen, H. P. 1987. Chem.

Rev. In press

73. Sharma, K. S., Sharma, K. 1980. Chromosome Techniques. Boston: Butterworth. 711 pp.

74. Sheppard, C. J. R., Wilson, T. 1982.

Proc. R. Soc. London Ser. A 379: 145

75. Shurcliffe, W. A. 1962. Polarized Light.

Cambridge, Mass: Harvard Univ. Press. 207 pp.

76. Singham, S. B., Salzman, G. C. 1986. J.

Chem. Phys. 84: 2658

77. Sipski, M. L., Wagner, T. E. 1977. Bioi.

Reprod. 16: 428

78. Thompson, R. C., Bottiger, J. R., Fry, E. S. 1980. Appl. Opt. 8: 1323

79. Thompson, R. C., Fry, E. S., Bottiger, J.

R. 1977. Proc. Soc. Photo-Opt. Instrum. Eng. 112: 152

80. Tinoco, I. Jr. 1959. J. Am. Chem. Soc. 81: 1540

81. Tinoco, I. Jr. 1979. Int. J. Quantum Chem. 16: 111

82. Tinoco, I. Jr., Bustamante, C., Maestre, M. F. 1980. Ann. Rev. Biophys. Bioeng. 9: 107

83. Tinoco, I. Jr., Ehrenberg, B., Steinberg, I. Z. 1977. J. Chem. Phys. 96: 916

84. Tinoco, I. Jr., Maestre, M. F., Bustamante, C. 1983. Trends Biochem. Sci. 8: 41

85. Tinoco, I. Jr., Turner, D. H. 1976. J.

Am. Chem. Soc. 98: 6453

86. Tinoco, I. Jr., Williams, A. L. Jr. 1984.

Ann. Rev. Phys. Chem. 35: 329

87. Tobias, I. 1984. Biopolymers 23: 1315

88. Tobias, I., Brocki, T. R., Balazs, N. S. 1975. J. Chem. Phys. 62: 4181

89. Turner, D. H. 1978. Methods Enzymol. 49G: 199

90. Turner, D. H., Tinoco, I. Jr., Maestre, M. F. 1974. J. Am. Chem. Soc. 96: 4340

91. Turner, D. H., Tinoco, I. Jr., Maestre, M. F. 1975. Biochemistry 14: 3794

92. van de Hulst, H. C. 1957. Light Scattering by Small Particles. New York:

Wiley. 470 pp.

93. Wells, K. S., Beach, D. A., Keller, D., Bustamante, C. 1986. Biopolymers 25: 2043

94. Wilson, T., Sheppard, C. 1984. Theory and Practice oj Scanning Optical Microscopy. New York: Academic. 213 pp.

95. Zeitz, S., Belmont, A., Nicolini, C. 1983.

Cell Biophys. 5: 163

You might also like