Looking at long molecules in solution: what happens when they are
subjected to Couette flow?w
Alison Rodger,aRachel Marrington,aMichael A. Geeves,bMatthew Hicks,ac
Lahari de Alwis,aDavid J. Halsalldand Timothy R. Daffornc
Received 3rd April 2006, Accepted 24th May 2006
First published as an Advance Article on the web 14th June 2006
Knowing the structure of a molecule is one of the keys to deducing its function in a biological
system. However, many biomacromolecules are not amenable to structural characterisation by the
powerful techniques often used namely NMR and X-ray diffraction because they are too large, or
too flexible or simply refuse to crystallize. Long molecules such as DNA and fibrous proteins are
two such classes of molecule. In this article the extent to which flow linear dichroism (LD) can be
used to characterise the structure and function of such molecules is reviewed. Consideration is
given to the issues of fluid dynamics and light scattering by such large molecules. A range of
applications of LD are reviewed including (i) fibrous proteins with particular attention being
given to actin; (ii) a far from comprehensive discussion of the use of LD for DNA and
DNA–ligand systems; (iii) LD for the kinetics of restriction digestion of circular supercoiled
DNA; and (iv) carbon nanotubes to illustrate that LD can be used on any long molecules with
accessible absorption transitions.
Many molecules in biological systems are long (e.g. DNA) or
part of long assemblies of molecules (e.g. fibrous proteins or
membrane-bound molecules). While it is clear that the struc-
tures of such moieties are important to their function, many of
the powerful techniques of structural biology, including X-ray
diffraction and NMR spectroscopy, are not well suited to high
order macromolecular complexes. Microscopies of various
kinds are extremely useful but if working on molecular scales,
almost by definition, focus on single units (molecules or arrays
of molecules) at any one time rather than the whole popula-
tion and the samples are also not fully in the solution phase.
The purpose of this article is to illustrate the advantages and
disadvantages of solution-phase flow linear dichroism (LD)
spectroscopy for structural characterization of solutions of
long molecules ranging from biomacromolecular systems to
carbon nanotubes. Our recent work in this area has been
inspired by the increased availability of samples for structural
analysis which in turn has led us to develop a range of new
Couette flow cells to reduce both sample volume and light
scattering and to enable experiments under tight temperature
control.1–5The way has now opened to many more applica-
tions that were previously impossible due to restricted sample
quantity, required conditions or molecule size.
Much of the flow linear dichroism literature relates to
nucleic acids. The DNA work published before the early
1990s has been superbly reviewed by Norde ´ n, Kubista and
Kuruscev in ref. 6. An earlier linear dichroism review by
Norde ´ n is more general in subject matter.7The aim of this
article is to cover the developments that have happened in the
last ten years or so and to give an overview of how linear
dichroism can be used to study a wide range of long molecules
from DNA to fibrous proteins to carbon nanotubes. The
intent here is to be illustrative rather than comprehensive,
though the references cited open up a literature trail that
should lead to most available literature.
Linear dichroism (LD) is a spectroscopic technique that can be
used with systems that are either intrinsically oriented, or can
be oriented during an experiment by external forces.6–8LD
measures the difference in absorption of light linearly po-
larised parallel and perpendicular to an orientation axis,
LD = AJ– A>. (1)
In the case of flow orientation of long molecules, the orienta-
tion axis is the long axis of the molecule. If a transition
moment (the direction of net electron displacement during
an electronic transition) is aligned more parallel than perpen-
dicular to the orientation axis, a positive LD signal is ob-
served. Conversely, if the direction of electron movement is
more perpendicular than parallel to the orientation axis, a
negative LD signal is observed. When a transition moment is
at an angle of 54.71 the LD signal equals zero, however
effective the orientation. Some transition polarizations of
relevance for biomacromolecules are shown in Fig. 1.2,9–11
aDepartment of Chemistry, University of Warwick, Coventry,
UK CV4 7AL
bDepartment of Biosciences, University of Kent, Canterbury,
UK CT2 7NJ
cDepartment of Biosciences, University of Birmingham, Birmingham,
UK B15 2TT
dDepartment of Clinical Biochemistry, Addenbrooke’s Hospital,
Cambridge, UK CB2 2QR
w The HTML version of this article has been enhanced with colour
This journal is ? c the Owner Societies 2006Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 | 3161
INVITED ARTICLE www.rsc.org/pccp | Physical Chemistry Chemical Physics
The reduced linear dichroism LDris often a convenient
pathlength- and concentration-independent summary of LD
data for quantitative analysis:
2Sð3cos2a ? 1Þð2Þ
where A is the absorption of the sample under isotropic
conditions i.e. not oriented, and S a scaling factor (the
orientation factor) defining the efficiency of the macroscopic
orientation. For uniaxial rods, S equals 1 for perfect orienta-
tion, and 0 for random orientation. a (Fig. 1) is defined as the
angle between the transition moment responsible for the
absorption of light at a particular wavelength and the orienta-
tion axis. If either S or a is known the other can be calculated.
Molecules can be aligned using a number of techni-
ques,6–8,12the most common being stretched film and flow
orientation, which is the focus of this article. Flow orientation
is ideal for biological molecules and others where the sample
needs to be hydrated.12–15The orientation of long polymers is
effected by the viscous drag created when a solution is flowed
between narrow walls. A Couette flow LD cell, where a
solution is placed in the annular gap between two cylindrical
cells, one of which rotates and causes alignment of the
molecules, has proved to be the most sample-efficient method
of achieving this. A schematic diagram of a Couette cell is
given in Fig. 2. For efficient alignment, the flow must be
laminar, not turbulent.
Couette cells are derived from the work of Maurice Fre ´ de ´ ric
Alfred Couette and Henry Reginald Arnulph Mallock, who in
the late 1800s independently developed a means to measure
viscosity based on shearing a liquid between coaxial cylin-
ders.16–18,26Since then there have been many applications of
the Couette principle within the engineering and physics
communities, the most notable example being Taylor–Couette
flow which is a term used to define the flow between rotating
The modern incarnation of the Couette system developed in
1964 by Wada and Kozawa20involves one cylinder being
inserted inside another and the solution flowing between the
narrow gap (annular gap) as it is dragged by the rotation of
one of the cylinders (Fig. 2). One of the cylinders must be
transparent to the radiation being used and the other needs a
transparent light path for the light which is incident radially
on the cell (Fig. 2). The annular gap has typically been 500 mm,
though experiments using a 50 mm annular gap have been
reported.13,15In order to orient the sample it is necessary for
the inner and outer cylinders to rotate at different speeds,
therefore, creating a viscous drag and flow gradient in the
solution; usually one cylinder is stationary. Previous studies
have used either a rotating inner and fixed outer cylinder e.g.
ref. 20 and 30, or a rotating outer and fixed inner cylinder e.g.
ref. 31 and 32. The latter offers more flow stability but there
are advantages for both options. In both cases it is necessary
(c) poly-proline type II helix (as determined by calculations2); (d) tryptophan;9(e) tyrosine; (f) adenine10and (g) guanine11chromophores together
with LD schematic.
The orientation of various polarization moments (a) a-helix (as determined by calculations2); (b) b-sheet (as determined by calculations2);
Fig. 2Schematic diagram of a Couette flow cell.
3162 | Phys. Chem. Chem. Phys., 2006, 8, 3161–3171This journal is ? c the Owner Societies 2006
that the light path through the cell allows optimal transmis-
sion, i.e. the rotating component and the region which is
stationary where the light beam is incident need to be trans-
parent to the radiation used. For example, for UV applications
quartz is the most common material (though calcium fluoride
has also been used15). The light beam can be either perpendi-
cular20or parallel31to the axis of rotation. The perpendicular
orientation is more common and is the one adopted in all
studies detailed here.
Linear dichroism spectra are usually collected using a con-
verted circular dichroism spectropolarimeter. The circularly
polarised light of a CD spectropolarimeter is converted to
linearly polarised light by the addition of a quarter wave plate,
or as in our experiments, the photo elastic modulator 1/4 wave
plate is converted to a 1/2 wave plate and the required
alternating polarizations of light are produced directly by
the instrument. Alternatively one can use polarisers in a
normal absorption spectrometer and manually rotate the
polarisers or the sample (though the latter is challenging for
a Couette flow cell). The former option has significantly better
signal to noise ratios. The linearly-polarised light passes
through the sample and interacts with the oriented molecules
to yield a net LD signal. As with all absorbance spectroscopy
techniques a baseline must be subtracted from the sample
spectrum. There are in principle two options for the baseline in
a flow LD experiment. The first is to stop (or almost stop) the
flow and measure the signal. With Couette flow cells this
requires that either the rotating unit is optically uniform or
that a very stable extremely low rotation speed is able to be
maintained. The former is the case for our recently designed
micro-volume Couette LD cells where the outer rotating
cylinder is an extruded quartz capillary. Alternatively, one
replaces the sample with buffer or solvent and rotates the unit
as for the sample.
Fluid dynamics of flow in concentric cylinder flow
A fluid is a substance that will deform continuously when it is
subjected to a tangential or shear force.33In the case of a
Couette system, a velocity is imposed on the fluid contained
between two concentric cylinders as one of these cylinders
rotates (Fig. 2). The viscosity of a fluid (its resistance to shear
or flow) will affect its dynamics. In all examples detailed in this
work the solutions are aqueous so can be assumed to act
predominantly like water, and therefore have constant viscos-
ity independent of shear gradient, as discovered by Couette26
(though high concentrations of sample increase the viscosity).
Different flow states within the system exist, and it is these
states that have been the subject of much study over the years
since the work of Mallock and Couette.16–19,26The most
notable work is that of Sir Geoffrey Ingram Taylor who
investigated the flow stability of a viscous fluid when two
cylinders rotate in the same direction and in opposite direc-
tions27and later went on to conclude that if only the outer
cylinder rotates the flow is more stable than if only the inner
cylinder rotates.28He also went on to study the effect of
rotation speed and investigate the effect this has on fluid
The spinning of either cylinder in a co-centric cylinder
system, as in the LD cells of Fig. 2, causes the fluid in the
annular gap to flow in a pattern dependent on many para-
meters in the system, including: the rotation speed of the
cylinder, which cylinder rotates, the viscosity of the fluid,
and the potential for interaction between the surface of the
cylinder and the fluid or molecules in the fluid.4,34The flow
profile for an inner rotating cylinder is given in Fig. 3.33
Transitions between states are determined as functions of the
inner and outer cylinder Reynolds numbers Riand Ro, respec-
tively. The Reynolds number is named after Osbourne Rey-
nolds35,36and is a dimensionless variable that indicates the
relative significance of the viscous effect compared to the
inertia effect. The Reynolds number (Re) in the form applic-
able for a fluid in the annulus between two concentric cylinders
is defined (kg s?1m?1)
where o1is the angular velocity of one rotating cylinder in rad
s?1, r1is the radius of that rotating cylinder in m, d is the gap
width (r2? r1) in m, and n is the kinematic viscosity in m2s?1
(kinematic viscosity is equal to absolute or dynamic viscosity
divided by density).37At high Re, the flow changes from
laminar Couette flow to a series of more complex states, the
first of which involves the appearance of Taylor vortices
(vertically stacked toroids of flow around the inner cylinder)
and the next involves wavy vortex flow which is exactly what
the name says. These can be visualised by introducing light-
reflecting suspended material into an LD cell as illustrated in
Fig. 4. In practice we found that the Taylor vortices and wavy
vortices could only be visualised in our original LD cell.12Our
more recent and better engineered cells proved to have much
more stable flow. Although Taylor vortex flow is still essen-
tially horizontal as in Couette flow, the slow twisting rotation
of the individual toroids is likely to have an effect on the
alignment of molecules and it may depend on their flexibility.
For example, we have observed that ethidium bromide bound
to DNA is differently oriented in Couette flow and Taylor
vortex flow regimes, presumably due to the different stiffness
of the DNA with and without the intercalator ethidium
from above. The longer the arrow describing the direction of the fluid
flow, the faster the flow speed. Figure reproduced from R. Marring-
ton, T. R. Dafforn, D. K. Halsall, M. Hicks and A. Rodger, Analyst,
2005, 130, pp. 1608–1616—Reproduced by permission of the Royal
Society of Chemistry.4
Fluid flow profiles for inner rotating cylinder set-up, viewed
This journal is ? c the Owner Societies 2006Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 | 3163
Light scattering correction
One of the challenges of working with high order macro-
molecular complexes, e.g. protein fibres or carbon nanotubes,
is that they scatter light. The lenses of the microvolume
capillary LD cells reduce this effect but it is still an issue, see
e.g. ref. 38 and 39. The problem arises because the spectro-
meter simply measures the light that reaches the detector and
assumes that which does not has been absorbed. Thus scat-
tered photons are assumed to have been absorbed. A signifi-
cant part of this problem can be removed by collecting the
light as close to the LD cuvette as possible—this can be
achieved by a number of methods including locating the
photomultiplier tube (PMT) very close to the sample, increas-
ing the angle of acceptance of the PMT, or locating a lens just
after the sample to gather the light onto the PMT. However,
even this does not always result in the desired flat line in the
LD spectra outside absorbance regions. In such cases, the
method of Nordh and others39can be used to subtract the
scattering contribution to the signal as follows. Absorbance
LD (LDA) and background turbidity dichroism (LDt) con-
tribute to the total LD signal as shown by eqn (5).
LDtotal= LDAþ LDt.(5)
The wavelength dependence of LDtis accounted for by linear
regression of eqn (5) to a function:
LDt(l) = al?k
where k is a constant that has been shown generally to relate to
the unpolarised turbidity (and usually takes values between 2.8
and 3.5), and a is a constant.39The implementation of this
method is illustrated in Fig. 5.
Applications of linear dichroism
In the 1960s–1970s flow LD was recognised as a key technique
and was used in the characterisation of fibrous proteins such
as actin as well as for probing the orientation of DNA and
bound ligands.21–24The limitations of instrument design
and technology then, in particular the sample volume required
(1–2 mL), and spectrometer design made it difficult to develop
the technique further. The main applications reported in the
literature lie in the analysis of nucleic acids,25mainly because
of the research interests of groups with Couette flow cells and
also because proteins can be difficult to orient (globular
proteins), or generate artefacts due to light scattering caused
by their large size. In the remainder of this article the applica-
tions of LD to a range of systems will be summarised. It
should be noted that the discussion of DNA applications is far
from complete due to space limitations.
The cytoskeleton of a cell is a network of protein fibres whose
repeat unit is a whole protein molecule and which form
microtubules and microfilaments in the cytoplasm. The cyto-
skeleton is critical to cell motility (cell movement) and cell
Groton, Massachusetts 01450, USA. http://www.kalliroscope.com) at a concentration of 5% in water in a concentric cylinder LD cell rotating at
different speeds (speed increases linearly with voltage). A light source was shone onto the cell from above to illuminate the flakes. The AQ-1000
flakes align along the direction of fluid flow, providing an insight into the actual fluid flow in the cell. A 1 cm high slice through the middle of the
cell window is illustrated at voltages from 2.6–3.5 V.
Images of AQ-1000 rheoscopic fluid (titanium dioxide coated mica flakes in silicone oil, Kalliroscope gallery, 264 Main Street, P.O. Box 60,
applied to a polymerised tubulin LD spectrum: the experimental data
(—) (from which a baseline of the same sample in a non-rotating cell
has been subtracted), the calculated turbidity LD, using a k value of
3.5, with a determined by rescaling the curve at 320 nm where there is
no intrinsic absorbance (- - -), and the corrected data is (– – –).
A plot to show the method of light scattering correction when
3164 | Phys. Chem. Chem. Phys., 2006, 8, 3161–3171This journal is ? c the Owner Societies 2006
morphology as well as processes such as cell division. It is not
a rigid permanent structure, it is dynamic and constantly
rearranges by polymerising and depolymerising to produce
movement. There are three major types of protein fibre within
the cytoskeleton of eukaryotic cells: actin filaments; intermedi-
ate filaments; and tubulin microtubules which are held to-
gether by protein–protein interactions. Apart from having
structural roles within the cell, both tubulin microtubules
and actin filaments also serve as ‘tracks’ for the biological
movement of motor proteins—dyneins and kinesins along
microtubules and myosins along actin filaments. More re-
cently, interest has grown in fibrous proteins implicated in
the mechanism of disease for such disorders as Alzheimer’s
and bovine spongiform encephalitis (prions). LD can be used
to study these systems as discussed below. Actin was the first
fibrous protein to be studied by Couette flow LD and recently
we have improved the quality of data one can collect for such a
system and also devised a range of new experiments. It is
therefore an ideal case study to show possibilities of LD of
fibrous proteins and is explored in some detail below. A brief
review of work on other protein fibres is given after the
discussion of actin.
Actin and actin-binding proteins. Actin exists as a globular
monomer called G-actin which polymerises spontaneously to
F-actin filaments in physiological conditions in the presence of
ions such as Mg21, K1or Na1. The process is reversible, in
that when the ionic strength of the solution is lowered F-actin
disassembles to G-actin monomers.40Actin binds 50-adenosine
triphosphate (ATP) or 50-adenosine diphosphate (ADP)41
which stabilise the fibre, but are not required for polymerisa-
tion. The three dimensional structure of actin molecules and
actin filaments was solved in 1990 by Kenneth Holmes,
Wolfgang Kabsch, and their colleagues.42
The polymerisation of actin has been a subject of interest for
many years. The most common methods of analysis include
viscometry, sedimentation, fluorescence spectroscopy, electron
microscopy and light scattering, all of which have their merits.
The initial application of Wada’s Couette LD instrument in
the 1960s and 1970s was to the study of F-actin solutions with
a view to understanding the mechanism of muscle contrac-
tion.21,24,43–45Higashi and Oosawa, using the apparatus devel-
oped by Wada and Kozawa in 1964,20were able to investigate
the orientation of aromatic amino acid residues (tyrosine and
tryptophan) and bound adenine nucleotides. They reported
that F-actin oriented by flow shows a negative dichroism at
about 260 nm (due to adenine), a positive one at about 280 nm
(tyrosine and tryptophan) and another negative one at about
295 nm (tryptophan).21
The use of LD to study actin did not proceed further at that
time as it was concluded that the sensitivity of the apparatus
needed to be improved to reduce light scattering and allow
data collection at lower wavelengths.45Recent developments
in spectrometer instrumentation and our new capillary
Couette cells provide the required improvement and as shown
below it is now possible to use LD as a technique either for,
e.g., the screening of actin–ligand binding or deciphering
structural characteristics of the protein.13
Fig. 6 shows a near UV LD spectrum of actin at 50 mM (see
Appendix for experimental methods) which is consistent with
those in the literature.46Far UV LD data for the backbone of
actin was first collected using LD by Dafforn et al. in 2004.13
The far UV LD spectrum of actin46at concentrations from
12–94 mM is shown in Fig. 7 and shows positive signals for
both the n - p* transitions and the higher energy component
of the p - p* transition, the lower energy component of the
p - p* is evident as the 215 nm dip in the spectra. The LD
therefore indicates that, on average, the a-helices in the fibre
are oriented more perpendicular than parallel to the fibre axis
in accord with literature data. Care must be taken in collecting
low wavelength data for a system such as actin where light
scattering is significant as illustrated in Fig. 7 where diluting
the sample leads to an apparent wavelength shift of the
maximum LD signal of B30 nm to shorter wavelengths. When
the dilution effect is simply to reduce the signal intensity
according to the Beer Lambert law, then we can conclude
the spectrum is real. From Fig. 8 it can be concluded that the
LD wavelength maximum of the higher energy component of
the p - p* transition is at 197.5 nm, in contrast to the
previously reported maximum at 200 nm.13In all such experi-
ments on fibrous proteins in general and actin in particular it is
important to investigate the stability of the formed fibres to
sheer forces by measuring the LD as a function of rotation
The monomeric G-actin has no orientation in the flow,
therefore the LD signal (shown in Fig. 7b at 222 nm) provides
a convenient signal to follow the process of polymerization.13
LD also provides a way to monitor the binding of globular
proteins to fibrous proteins since globular molecules will not
normally orientate in a flow cell, but upon binding to actin
they do. The myosin motor domain (or subfragment 1, S1)
binds tightly and stoichiometrically to F-actin. On its own, it
has no LD spectrum (Fig. 6) but when bound to actin its near
UV spectrum is distinct from the F-actin spectrum with loss of
the negative peaks at 295 and 275 nm and enhanced 285 and
290 nm positive peaks reflecting the differences in the average
(- - -) and actin (50 mM)-myosin S1 (50 mM) (– – –) in KCl (15 mM);
MgCl2(0.75 mM) and MOPS (3 mM) pH 7. The LD spectrum of
unlabelled actin (50 mM)–myosin S1 (50 mM) has been corrected for
baseline light scattering (k = 3.5, zeroing at 450 nm).
Near UV LD spectra of actin (50 mM) (—); myosin S1 (50 mM)
This journal is ? c the Owner Societies 2006Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 | 3165
orientations of aromatic residues in actin and the actin–
Tropomyosin is another actin-binding protein, but in this
case it is itself a linear molecule and it binds to actin with a low
stoichiometry of A7Tm. Tropomyosin (Tm) is a coiled-coil
protein that aligns in Couette flow and has its own LD
spectrum, as first reported in ref. 47 where data were collected
with a large-volume LD cell down to B210 nm. The LD
spectrum of Tm (Fig. 9) has a broad negative LD signal
between 218–240 nm (n - p* transition) and a positive LD
maxima at 206 nm.46Below 206 nm the LD signal tends
negative. At higher concentrations a positive LD signal in
the near UV at 280 nm due to the tyrosines (Tm contains no
tryptophans) is apparent. Tm is a highly helical protein
(>90% a-helices48), and the n - p* transition of the a-helices
is perpendicular to the orientation axis, so its negative LD is
consistent with the a-helices being more parallel than perpen-
dicular to the orientation axis, as one would expect for a
coiled-coil. The presence of Tm bound to actin has little effect
on the LD spectrum of actin, largely because of the weak Tm
signal. Tm forms a 1 : 1 complex with troponin (Tn, a
globular protein which therefore has no intrinsic flow LD
signal) which then binds to actin in a 7 : 1 : 1 ratio. Fig. 10
shows LD spectra of phalloidin-stabilised actin (psA, to make
extended data collection feasible), Tm and Tn individually,
psA-Tm and psA-Tn mixtures and in a 1 : 1 : 1 mixture to
ensure saturation of actin with Tm and Tn. Tn increases the
psA signal slightly, Tm shifts its maximum wavelength and
also leads to a slightly increased signal, whereas the 1 : 1 : 1
mixture spectrum has both a shift of wavelength and a 30%
larger LD signal. These data indicate that Tn’s main effect is
on the orientation factor of the fibre. Upon addition of CaCl2,
the LD spectrum (of psA–Tm–Tn–Ca21) overlays that of
psA–Tm consistent with calcium causing the Tn–actin inter-
action to weaken (data not shown). Near UV LD studies by
Yanagida et al. in 1974 on the effect of Ca21on the F-
actin–Tm–Tn complex showed about a 20% decrease in
intensity of the two negative peaks at 260 and 295 nm, and
concluded that the presence of calcium made the fibre more
flexible.24,45Taniguchi reported an even larger decrease in LD
intensity upon the addition of calcium ions.
Near UV LD of actin (21 mM) and its complex with Tm
(3 mM), Tm–Tn (3 mM) and Tm–Tn–Ca21(Tm–Tn 3 mM and
Ca213 mM) are shown in Fig. 11.46It can be seen that there is
an increase in signal intensity of the 285 nm transition
(tyrosine and tryptophan) upon binding of Tm, and even
F-actin concentrations 93; 74; 62; 53 and 12 mM (the true spectrum, solid line). (b) Polymerisation of G-actin into F-actin (B0.4 mg mL?1) with
ATP present (a) kinetics analysis monitoring LD222nm.
(a) Far UV LD spectra showing the apparent shift to shorter wavelength of the maximum signal as the concentration of F-actin is reduced.
concentrations of 1.2 (...); 1.0 (.....); 0.8 (– – –); 0.6 ( - - - -); 0.4 (? ?– –) and 0.2 mM (—?), and (b) linear plots showing that the Beer Lambert law
is obeyed at the concentrations shown.
The concentration dependence of the LD of solutions of F-actin at low concentrations, (a) wavelength LD spectra with F-actin
3166 | Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 This journal is ? c the Owner Societies 2006
more so (by B30%) upon binding of the Tm–Tn complex. The
ADP region (265 nm) shows a slight increase upon addition of
Tm to actin, with no noticeable change recorded for the
addition of Tm–Tn to actin. There is no obvious difference
in the tryptophan signal at 295 nm except for the binding of
the Tm–Tn complex to actin. The effect of the addition of
calcium to the system reduced all three signals by B30%. This
is consistent with literature data, though no change in the
285 nm transition had previously been reported. This may be
due to the methods employed for the correction of scattering
when these analyses were originally undertaken. As concen-
trations of greater than 100 mM were used in 1970s, light
scattering artefacts were very significant in the spectra col-
When using linear dichroism to detect the binding between a
fibrous protein and globular protein (such as actin and
myosin) it can be difficult to determine whether changes in
the LD are due to the globular protein once it is oriented and/
or whether there is a change in the orientation of the fibre as a
whole or even its subunits. To aid in the interpretation of such
spectra a probe molecule attached to the fibre that allows the
orientation factor to be evaluated is required. The ideal probe
has yet to be found, but Fig. 12 shows that probes can be
readily detected. Rhodamine labelled phalloidin is attractive
as the phalloidin stabilises the actin filaments without affecting
the structure (rhodamine-labelled phalloidin stabilised actin
and actin have very similar shaped LD spectra in the UV
despite contributions from the ligand) and rhodamine has a
large extinction coefficient (106 000 M?1cm?1at 542.75 nm49)
and the signals are linear with concentration.
Far UV LD of other fibrous proteins
As noted above, actin was the first protein fibre to which
Couette flow LD was applied in the near UV wavelength
region and it is now possible to collect far UV data on such
a system. A similar story can be told for tubulin (where a,b-
tubulin heterodimers nucleate into oligomers, which then form
protofilaments, with the subunits aligned head to tail with
dimeric units repeating every 8 nm). Early experiments probed
the near UV38or used the turbidity LD signal outside the
absorbance region to follow polymerisation.39Backbone data
for tubulin have more recently been collected and used to
monitor chromophore reorientation during the polymerisation
process.5One of the challenges of tubulin is the need to hold
the sample at 37 1C during polymerisation. Analysis of tubulin
kinetics and binding of various ligands including Taxolt,
colchicine, vincristine and DAPI have been probed in a
microvolume Couette cell.5
Essentially any fibrous protein which stays in solution can
be flow oriented; at least preliminary near and far UV LD data
have been collected for prions,50Alzheimer’s fibres, collagen,
and a1-antitrypsin.11LD has also been used to probe the
processes involved in the polymerisation and bundling of the
bacterial homologue of tubulin, FtsZ.51The E. coli FtsZ has
no tryptophans so the near UV region of the spectrum can be
(– – –); psA–Tm (----); psA–Tm–Tn (.....). All protein concentrations
are 1 mM. All solutions contain KCl (15 mM); MgCl2(0.75 mM) and
MOPS (3 mM) pH 7.
(a) LD of phalloidin stabilised actin (psA) (–––), psA–Tn
KCl (20 mM), MgCl2(1 mM) and MOPS (4 mM) pH 7.
LD spectra of tropomyosin at 50 (–––) and 10 mM (- - - -) in
mM)–tropomyosin (Tm) (3 mM) (.....); A (21 mM)–Tm (3 mM)–
Troponin (Tn) (3 mM) (----) and A (21 mM)–Tm (3 mM)–Tn
(3 mM)–Ca21(3 mM) (–––). All solutions contained KCl (15 mM);
MgCl2(0.75 mM) and MOPS (3 mM) pH 7.
Near UV LD spectra of actin (A) (21 mM) (–––); A (21
This journal is ? c the Owner Societies 2006Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 | 3167
interpreted in terms of contributions from guanine and tyr-
osines. This has been used to follow some of the structural
changes which occur when the protofilaments bundle to form
fibres. LD can also be used to probe the binding of another
protein to a fibre. For example, the E. coli ortholog of the
accessory protein Zap A, YgfE (which has been shown to be a
bonafide division protein52) has been shown to promote FtsZ
bundling at physiological concentrations.52During the bund-
ling process LD shows that the guanine in GTP (which is
sandwiched between successive FtsZ units) changes orienta-
tion. This has led to the proposal that macrochelation by a
divalent metal cation to the terminal phosphate of GTP and
the N7 nitrogen of guanine causes the conformational change
of the guanine. However, whether this initiates or results from
FtsZ polymer bundling is at present unknown.
DNA and DNA–ligand systems
An article with the chosen title would not be complete without
at least some mention of DNA and DNA–ligand systems, due
to the extensive literature available in this area. However, this
also presents a problem as it is very difficult to give a
representative view of the whole field. B-DNA is a long
molecule whose base p–p* transitions all lie perpendicular to
the helix axis in an idealised structure—thus giving rise to a
negative signal under the DNA absorbance bands with a
maximum magnitude at the same wavelength (almost) as the
normal absorbance maximum. The consensus is that rather
than 901, a value of 861 should be used (though the ‘true’ value
is probably less than this and depends on the flexibility (so
environment and sequence) of a given piece of DNA). Chou
and Johnson53extensively analysed the LD of DNA down to
175 nm in terms of the base transitions and found the average
inclination of the bases is of the order of 201 from the
One of the main applications of LD uses its ability to probe
interactions between an oriented molecule and a species that
can not be aligned by itself but aligns upon binding to the long
molecule. Information regarding the corresponding binding
geometries can be ascertained if transition moment directions
of the components are known, or conversely. Molecules
(usually cationic species) can bind to DNA by intercalation,
groove binding or externally binding. Examples include dyes,
metal complexes and proteins. There are a number of superb
reviews on this area, as noted above.6,7The work of Schellman
and collaborators and Norde ´ n and collaborators has been
particularly valuable. Table 1 gives examples of DNA–ligand
complexes studied by LD.
When ligands are added to a solution of DNA and they
bind, one expects to see signals due to the ligand transitions as
well as the DNA. Ideally the ligand has no absorbance in the
DNA region, so one can determine the DNA orientation
parameter in the presence of the ligand. Intercalators are
planar aromatic molecules and bind parallel to the DNA
bases, so to a first approximation are expected to have LDr
values the same as the DNA bases. In practise, they locally
stiffen the DNA so they and the bases near them are better
oriented than other bases and the LDris greater in the ligand
region.8Thus, if all ligand transitions have negative LD signals
with LDrvalues greater (due to increase in S) or equal in
magnitude to those of the DNA bases, then one can be fairly
sure that the ligand is intercalating. Alternatively, if long-axis
polarised ligand transitions have positive LD signals, then a
groove-binding mode is indicated. If there is a preference for
AT-rich regions of DNA (where the groove is less sterically
hindered) this supports this binding mode assignment.54If S is
known, one can be more quantitative above this conclusion. If
(– – –), 4 (---) and 2 mM (? ? – –) showing linear increase of LD signal with concentration.
(a) LD of actin (50 mM) (–––) and pyrene labelled actin (50 mM) (....). (b) LD of rhodamine phalloidin labelled actin 10 (–––), 8 (....), 6
table together with ref. 6 and 7 gives the beginning of a literature trail
to find most systems which have been studied
Examples of DNA–ligand complexes studied by LD. This
Type of compound
bound to DNA Ligand bound to DNARef.
Dyes Ethidium Bromide
Ruthenium metal complexes
Di-iron metal complexes
3168 | Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 This journal is ? c the Owner Societies 2006
the only available ligand transition is in the DNA region one
either has to hope S has not changed or hope that there is a
wavelength where DNA can be probed independently of the
ligand, as was the case for anthracene deriviatised with
cationic spermine.55In this case a slight bending or stiffening
of the DNA was observed. In the case of a tetracationic major
groove binding di-iron helicate, the DNA was significantly
bent locally to the ligand binding site, so the ligand orientation
on the DNA cannot be determined. However, at low loading
the bending per ligand was calculated.56An additional chal-
lenge in quantifying ligand orientation is that only bound
ligands contribute to the LD signal but unbound ones con-
tribute to the absorbance, thus making LDrdeterminations
challenging if the binding constant and the different extinction
coefficients of free and bound ligand are not known.57
The extent of orientation of a DNA molecule depends on its
flexibility and also its length. Simonson and Kubista68empiri-
cally related LD intensity to DNA length on the basis of the
LDrof some long DNAs of well-defined length:
where A is the absorbance, G is the velocity gradient of the
flow cell, k1= 0.42 for our experiment (from Fig. 8 of ref. 68),
and the inverse of 1/k2is proportional to the DNA length.
This equation has since been used to quantify the bending
induced in DNA upon addition of a ligand.56The minimum
length of DNA required to give an LD signal depends on the
DNA flexibility and the desired signal : noise ratio. For easy
data collection, 1000 or more base pairs are desirable. Less
than 250 base pairs will be hard work—not surprisingly given
that the persistence length (the length of DNA for it to bend
though 1 rad) of DNA is typically B150 base pairs. Less
quantitatively, changes in LD can be used to monitor single
nucleotide polymorphisms since DNA mis-matches have
an effect on DNA shape (hence scattering) and DNA
Using molecular length to follow kinetics
LD is the ideal method to probe the kinetics of formation of
protein fibres from assemblies of monomeric units as discussed
above. It is also the ideal technique to follow such reactions as
the effect of a restriction enzyme (DNA-cutting enzyme) as a
function of time.70As illustrated in Fig. 13 if one begins with
comparatively compact supercoiled circular DNA and adds a
restriction enzyme the LD signal increases as the average
length and hence orientation of the DNA increases. If the
enzyme has two cut sites, then a subsequent decrease in LD is
noticed as the average length decreases again. Perhaps the
most intriguing thing about this experiment is that a super-
coiled plasmid DNA (in this case B7000 base pairs in length)
can be readily oriented—the starting point is not a zero LD
signal. It is conceivable that, coupled with appropriate model-
ling, LD may be able to give information about the structures
adopted by such a molecule.
LD is also ideal for following the progress of polymerase
chain reactions (PCRs) or even simply monitoring the end
point of a reaction.71In the latter case it can be used to
determine viral load and by careful choice of primers this can
be selective. The advantage of LD over other real time PCR
methods is that primer-dimers, which plague most quantitative
PCR methods, are invisible. Also, no additional labelling is
Carbon nanotubes provide a final example for this article and
‘proof’ that one can use flow LD for any long molecule which
can be put into solution and is spectroscopically active. Single
walled carbon nanotubes (SWNTs) are soluble (more or less)
in aqueous SDS. Despite the size of the SWNTs the scattering
does not dominate the LD signal and the nanotube itself has a
broad band in the far UV region of the spectrum.72,73Any
small molecule which binds to DNA also acquires an LD
signal, though in the case of aromatic molecules it is clear that
there is significant coupling between the graphene sheets of the
nanotubes and the p-systems of the ligand as well as inter-
ligand interaction (as shown by the concentration dependence
of the shape of the spectrum in Fig. 14).70Despite the negative
charge of DNA we have also oriented DNA and the neutral
PNA on SWNTs. The signs and magnitudes of the DNA
signals should enable an accurate average orientation of the
DNA on the SWNT to be determined if we know the orienta-
tion parameter. So far only qualitative analyses have been
possible which indicate that DNA wraps around the nanotube
in the opposite manner from PNA.73
Other recent applications of LD include the analysis of
proteins and peptides in lipid bilayers, for example cyto-
chrome c inserted into soybean liposome,15and the orienta-
tion of peptide fibres that are designed for molecular self
assembly.74LD has also successfully been used to provide
experimental evidence for models postulated by computer
plasmid with EcoRI (which has a single cut site in the DNA sequence
and thus single step kinetics) and EagI (which has two cut sites in the
Kinetics of restriction digests of a circular, super-coiled
This journal is ? c the Owner Societies 2006Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 | 3169
simulations.64An ongoing project is the use of computer
modelling programs for further structural characterisation of
LD spectra of protein fibres.15So in conclusion it really does
seem as if flow linear dichroism can be used to study any long
molecule whose absorption spectrum is accessible.
Appendix: Materials and methods for actin studies
Unlabelled and pyrene-labelled actin were prepared using the
methods detailed in ref. 75 and 76. Phalloidin-stabilised
F-actin was prepared by incubating a solution of 10 mM actin
(unlabelled or labelled) with 10 mM phalloidin for at least one
hour at 4 1C.77Rhodamine phalloidin (R-415) was purchased
from Molecular Probes. Using the information provided on
the data sheet, a concentration of B200 mM was prepared by
dissolving the contents of the vial in 49.5 mL methanol
(spectroscopic grade). Rhodamine-labelled phalloidin-stabi-
lised actin was prepared by mixing rhodamine-labelled phal-
loidin and unlabelled actin in a 1 : 1 ratio at 10 mM for at least
one hour at 4 1C. Tropomyosin (Tm) and troponin (Tn) were
purchased from Sigma as lyophilized powder and solutions
were prepared on the day of analysis. A relative molecular
mass of B65 kDa was assumed for Tm (calculated using
Swiss-Prot accession numbers P04268 and P19352 for the a
and b chains of tropomyosin from chicken78) with a purity of
99% (as stated on the container). A relative molecular mass of
B73.0 kDa was assumed for Tn (using Swiss-Prot accession
numbers P02644; P12620 and P09860 for each subunit of
Tn78), with a purity of 91% (as stated on container).
1 R. Marrington, T. R. Dafforn, D. J. Halsall and A. Rodger,
Biophys. J., 2004, 87, 2002–2012.
2 A. Rodger, J. Rajendra, R. Marrington, M. Ardhammar, B.
Norde ´ n, J. D. Hirst, A. T. B. Gilbert, T. R. Dafforn, D. J. Halsall,
C. A. Woolhead, C. Robinson, T. J. Pinheiro, J. Kazlauskaite, M.
Seymour, N. Perez and M. J. Hannon, Phys. Chem. Chem. Phys.,
2002, 4, 4051–4057.
3 A. Rodger, J. Rajendra, R. Marrington, R. Mortimer, T. Andrews,
J. B. Hirst, A. T. B. Gilbert, D. Halsall, T. Dafforn, M. Ardham-
mar, B. Norde ´ n, C. A. Woolhead, C. Robinson, T. Pinheiro, K. J.
M. Seymour, N. Perez and M. J. Hannon, in Biophysical Chem-
istry: Membranes and Proteins, ed. R. H. Templer and R. Leather-
barrow, The Royal Society of Chemistry, Cambridge, 2002,
4 R. Marrington, T. R. Dafforn, D. J. Halsall, M. Hicks and A.
Rodger, Analyst, 2005, 130, 1608–1616.
5 R. Marrington, M. Seymour and A. Rodger, Chirality, 2006.
6 B. Norde ´ n, M. Kubista and T. Kuruscev, Q. Rev. Biophys., 1992,
7 B. Norde ´ n, Appl. Spectrosc. Rev., 1978, 14, 157–248.
8 A. Rodger and B. Norde ´ n, Circular Dichroism and Linear Dichro-
ism, Oxford University Press, Oxford, 1997.
9 B. Albinsson and B. Norde ´ n, J. Phys. Chem., 1992, 96, 6204–6212.
10 A. Holmen, A. Broo, B. Albinsson and B. Norden, J. Am. Chem.
Soc., 1997, 119, 12240–12250.
11 T. R. Dafforn, J. Rajendra, D. J. Halsall, L. C. Serpell and A.
Rodger, Biophys. J., 2004, 86, 404–410.
12 A. Rodger, in Methods in Enzymology, ed. J. F. Riordan and B. L.
Vallee, Academic Press, San Diego, 1993, vol. 226, pp. 232–258.
13 T. R. Dafforn, J. Rajendra, D. J. Halsall, L. C. Serpell and A.
Rodger, Biophys. J., 2004, 86, 404–410.
14 L. B. A. Johansson and A. Davidsson, J. Chem. Soc., Faraday
Trans. 1, 1985, 81, 1375–1388.
15 A. Rodger, J. Rajendra, R. Marrington, M. Ardhammar, B.
Norden, J. D. Hirst, A. T. B. Gilbert, T. R. Dafforn, D. J. Halsall,
C. A. Woolhead, C. Robinson, T. J. T. Pinheiro, J. Kazlauskaite,
M. Seymour, N. Perez and M. J. Hannon, Phys. Chem. Chem.
Phys., 2002, 4, 4051–4057.
16 M. Couette, Ann. Chim. Phys, 1890, 6, 433–510.
17 A. Mallock, Proc. R. Soc. London, 1888, 45, 126.
18 A. Mallock, Philos. Trans. R. Soc. London, Ser. A, 1896, 187, 41.
19 M. Kasai and F. Oosawa, Methods Enzymol., 1972, 25, 289–323.
20 A. Wada and S. Kozawa, J. Polym. Sci., Part A, 1964, 2, 853–864.
21 S. Higashi, M. Kasai, F. Oosawa and A. Wada, J. Mol. Biol., 1963,
22 J. Hofricheter and W. Eaton, Annu. Rev. Biophys. Bioeng., 1976, 5,
23 B. Norde ´ n, Appl. Spectrosc. Rev., 1978, 14, 157–248.
24 F. Oosawa, Y. Maeda, S. Fujime, S. Ishiwata, T. Yanagida and M.
Taniguchi, J. Mechanochem. Cell Motil., 1977, 4, 63–78.
25 B. Norde ´ n, M. Kubista and T. Kurucsev, Q. Rev. Biophys., 1992,
26 R. J. Donnelly, Phys. Today, 1991, November, 32–39.
27 G. I. Taylor, Proc. R. Soc. London, Ser. A, 1923, 223, 289–343.
28 G. I. Taylor, Proc. R. Soc. London, Ser. A, 1936, 157, 546–564.
29 G. I. Taylor, Proc. R. Soc. London, Ser. A, 1936, 157, 565–578.
30 A. Wada, Biopolymers, 1964, 2, 361–380.
31 C. Lee and N. Davidson, Biopolymers, 1968, 6, 531–550.
32 P. Oriel and J. Schellman, Biopolymers, 1966, 4, 469–494.
33 J. O. Wilkes, Fluid Mechanics for Chemical Engineers, Prentice Hall
PTR, Upper Saddle River, 1999.
34 C. D. Anderick, S. S. Liu and H. L. Swinney, J. Fluid Mech., 1986,
35 O. Reynolds, Philos. Trans. R. Soc. London, 1883, 174, 935–982.
36 O. Reynolds, Proc. R. Soc. London, 1883, 35, 84–99.
37 M. C. Potter and D. C. Wiggert, Mechanics of Fluids, Brooks/Cole
Thomson Learning, Pacific Grove, 3rd edn, 2002.
SWNTs (0.1 mg mL?1) in SDS (9 mM) solution. SWNTs–naphthalene
complexes were prepared by adding different amounts (0.2, 0.4 and 0.6
mg) of methanol dissolved naphthalene to 1 mL of SWNTs (0.1 mg
mL?1) in SDS (9 mM) solution and making final naphthalene con-
centrations of 0.2, 0.4 and 0.6 mg mL?1. (b) Difference spectra
obtained by subtracting the LD spectrum of SWNT (0.1 mg mL?1)
from the LD spectra of 0.2, 0.4 and 0.6 mg mL?1naphthalene–
(a) LD spectra of different concentrations of naphthalene in
3170 | Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 This journal is ? c the Owner Societies 2006
38 M. Taniguchi and R. Kuriyama, Biochim. Biophys. Acta, 1978,
39 J. Nordh, J. Deinum and B. Norden, Eur. Biophys. J., 1986, 14,
40 E. D. Korn, M. F. Carlier and D. Pantaloni, Science, 1987, 238.
41 S. L. Brenner and E. D. Korn, J. Biol. Chem., 1984, 259,
42 K. C. Holmes, D. Popp, W. Gebhard and W. Kabsch, Nature,
1990, 347, 44–49.
43 M. Miki and K. Mihashi, Biophys. Chem., 1977, 6, 101–106.
44 M. Taniguchi, Electro–Opt. Dielectr. Macromol. Colloids, 1979,
45 T. Yanagida, M. Taniguchi and F. Oosawa, J. Mol. Biol., 1974, 90,
46 R. Marrington, PhD Thesis, University of Warwick (Coventry),
47 M. J. Pandya, G. M. Spooner, M. Sunde, J. R. Thorpe, A. Rodger
and D. N. Woolfson, Biochemistry, 2000, 39, 8728–8734.
48 F. G. Whitby and G. N. Phillips, Jr, Proteins: Struct., Funct.,
Genet., 2000, 38, 49–59.
49 Molecular Probes, UK, 1993.
50 L. C. Serpel, 2003, unpublished data.
51 R. Marrington, E. Small, A. Rodger, T. R. Dafforn and S.
Addinall, J. Biol. Chem., 2004, 47, 48821–48829.
52 F. J. Gueiros-Filho and R. Losick, Genes Dev., 2002, 16,
53 P.-J. Chou and W. C. J. Johnson, J. Am. Chem. Soc., 1993, 115,
54 B. Norde ` n, S. Eriksson, S. K. Kim, M. Kubista, R. Lyng and B.
Akerman, in The Jerusalem Symposia on Quantum Chemistry and
Biochemistry, ed. B. Pullman and J. Jornter, Kluwer Academic
Publishers, Dordrecht, 1990, vol. 23, pp. 23–41.
55 A. Rodger, I. S. Blagbrough, G. Adlam and M. L. Carpenter,
Biopolymers, 1994, 34, 1583–1593.
56 M. J. Hannon, V. Moreno, M. J. Prieto, E. Molderheim, E.
Sletten, I. Meistermann, C. J. Isaac, K. J. Sanders and A. Rodger,
Angew. Chem., Int. Ed., 2001, 40, 879–884.
57 D. Z. Coggan, I. S. Haworth, P. J. Bates, A. Robinson and A.
Rodger, Inorg. Chem., 1999, 38, 4486–4497.
58 C. Houssier, B. Hardy and E. Fredercq, Biopolymers, 1974, 13,
59 M. Kubista, B. Akerman and B. Norden, Biochemistry, 1987, 26,
60 J. H. Moon, S. K. Kim, U. Sehlstedt, A. Rodger and B. Norden,
Biopolymers, 1996, 38, 593–606.
61 M. A. Ismail, K. Sanders, G. C. Fennell, H. C. Latham, P.
Wormell and A. Rodger, Biopolymers, 1998, 46, 127–143.
62 B. Norde ´ n and F. Tjerneld, FEBS Lett., 1976, 67, 368–370.
63 K. K. Patel, E. A. Plummer, M. Darwish, A. Rodger and M. J.
Hannon, J. Inorg. Biochem., 2002, 91, 220–229.
64 D. Z. M. Coggan, I. S. Haworth, P. J. Bates, A. Robinson and A.
Rodger, Inorg. Chem., 1999, 38, 4486–4497.
65 A. Rodger, A. Parkinson and S. Best, Eur. J. Inorg. Chem., 2001, 9,
66 F. Tjerneld, B. Norde ´ n and H. Wallin, Biopolymers, 1982, 21,
67 B. Norde ´ n, K. Mortensen, C. Elvingson, M. Kubista, B.
Sjo ¨ berg, M. Ryberg and M. Takahashi, J. Mol. Biol., 1991, 226,
68 T. Simonson and M. Kubista, Biopolymers, 1993, 33, 1225–1235.
69 T. R. Dafforn, D. J. Halsall and A. Rodger, Chem. Commun., 2001,
70 M. R. Hicks, A. Rodger, C. M. Thomas, S. Blatt and T. R.
Dafforn, Biochemistry, 2006, submitted for publication.
71 D. J. Halsall, T. R. Dafforn, R. Marrington, E. Halligan and A.
Rodger, IVD Technol., 2004, 6, 51–60.
72 J. Rajendra, M. Baxendale, L. G. Dit Rap and A. Rodger, J. Am.
Chem. Soc., 2004, 126, 11182–11188.
73 J. Rajendra and A. Rodger, Chem.–Eur. J., 2005, 11, 4841–4848.
74 M. J. Pandya, G. M. Spooner, M. Sunde, J. R. Thorpe, A. Rodger
and D. N. Woolfson, Biochemistry, 2000, 39, 8728–8734.
75 S. S. Lehrer and G. Kerwar, Biochemistry, 1972, 11, 1211–1217.
76 A. H. Criddle, M. A. Geeves and T. E. Jeffries, Biochem. J., 1985,
77 R. Maytum, S. S. Lehrer and M. A. Geeves, Biochemistry, 1999,
78 E. Gasteiger, A. Gattiker, C. Hoogland, I. Ivanyi, R. D. Appel and
A. Bairoch, Nucleic Acids Res., 2003, 31, 3784–3788.
This journal is ? c the Owner Societies 2006Phys. Chem. Chem. Phys., 2006, 8, 3161–3171 | 3171