Circular Dichroism, Applications
Giuliano Siligardi and Rohanah Hussain, Diamond Light Source Ltd., Diamond House, Didcot, Oxfordshire, UK
2017 Elsevier Ltd. All rights reserved.
Symbols
DA differential absorbance
AL absorbance from left-circularly polarized light
AR absorbance from right-circularly polarized light
[HT] total concentration of a host
[LT] total concentration of a ligand
[H] concentration of a bound host
Abbreviations
CD Circular dichroism (CD)
ECD electronic CD
HL hostligand complex
ITC isothermal titration calorimetry
Introduction
The most important and frequently used techniques to study,
monitor, characterize, and analyze binding interactions of
nonimmobilized molecules in solution are isothermal titration
calorimetry (ITC), fluorescence, and circular dichroism (CD).
For a detailed and thorough study of complex biosystems, one
technique is seldom enough. This is also because each tech-
nique has its own limitations.
ITC requires a significant amount of sample, particularly,
when the heat released upon binding is rather small. For more
complex systems where multiequilibria may exist, for instance,
intra- and intermolecular interactions (association), the bind-
ing can be very difficult to be determined correctly particularly
when the ligand and the host are in different solvents due to
solubility problems.
Fluorescence relies on the presence of fluorophores with
sufficient intrinsic fluorescence. The majority of proteins con-
tain useful fluorophores such as tryptophan and tyrosine
amino acid residues that can be used to probe binding inter-
actions. However, their intrinsic fluorescence is rather modest.
The introduction of more sensitive fluorophore probes by
chemical tagging might induce positive or negative results by
promoting or hindering the binding interactions. This is
encountered in high-throughput drug screening where the reli-
ability of positive and negative results to select lead com-
pounds cannot be dismissed. The open question is how good
is the screening test, in other words how many positive results
are biased by the artificial binding promoted by the tagged
fluorophore and how many bindings are missed, that is, neg-
ative results, induced by the steric hindrance of the tag moiety?
CD spectroscopy has none of these limitations; however,
the setting of the optimal instrumental and experimental
parameters to obtain the necessary good signal-to-noise CD
Encyclopedia of Spectroscopy and Spectrometry, Third Edition http://dx.doi.org/10.1016/B978-0-12-803224-4.00031-5
[L] concentration of a bound ligand
[HL] concentration of a bound hostligand complex
K equilibrium binding constant
D differential molar extinction coefficient
molar extinction coefficient
c concentration in mol l
1
l cuvette pathlength in cm
OR optical rotation
ORD optical rotatory dispersion
ROA Raman optical activity
VCD vibrational CD
spectra is less user-friendly than the other techniques. The
analysis of the data is not yet available as a comprehensive
software package like for ITC.
The golden rules on how to carry out good ligand titrations
and what proteinligand concentrations to choose in order to
study binding interactions of biologically important molecules
by CD spectroscopy are described in here.
Optical Activity
A chiral molecule is not superimposable on its mirror image
and as a consequence it shows optical activity in the form of
optical rotation (OR), optical rotatory dispersion (ORD), CD,
and Raman optical activity (ROA).
Linearly polarized light can be seen as the sum of two
circularly polarized lights in opposite directions, one left and
one right.
OR is the difference in speed between the propagation of
left- and right-circularly polarized lights that pass through a
chiral medium. With no light absorption from the sample, the
result is the rotation of the plane of linearly polarized light.
ORD is the change of OR as a function of wavelength.
CD is a small difference in the absorption between left- and
right-circularly polarized lights of a chiral molecule. Electronic
CD (ECD) is generated by molecules with chromophores
absorbing in the UV and visible spectral regions, whereas
vibrational CD (VCD) is generated in the IR spectral region.
ROA is a small difference in Raman intensity between right-
and left-circularly polarized lights or scattered circularly polar-
ized light of a chiral molecule.
Both CD and ROA arise from the absolute configuration
(the chirality, or handedness) and the molecular structure (the
conformation).
293
294 Circular Dichroism, Applications

The absolute configuration of small, relatively rigid mole- of ligand in a 20 ml volume, the concentration of the stock
cules with up to 30 atoms can now be determined, thanks to solution to be prepared is approximately 50-fold that of the
the revolutionary developments of density functional theory concentration of the host component.
methods for the calculation of ECD, VCD, and ROA spectra. The quadratic equation to calculate Kd has been derived
Rather demanding calculations including solvent interactions from a two-component system where only one component
for larger and more flexible molecules (peptides) with about has intrinsic CD (Figure 1). However, when both host and
120 atoms have been recently achieved. It is conceivable that in ligand components possess intrinsic CD (Figure 2), eqn [1]
a not distant future, maybe in the next decade thanks to less
computationally expensive calculations, small proteins will be
tackled leading the way to even larger protein systems. The
0
implication is that the 3D structure of proteins in solution
P2
could be determined one day with sufficient resolution from
observed and calculated ECD, VCD, and ROA spectra from a
1:6
pool of conformations sampled from quantum mechanical 1:3
molecular dynamics. 1:2
For biologically important molecules, such as proteins, 1:1.5
nucleic acids, carbohydrates, and lipids, CD spectroscopy is 1:1
1:0.6
the technique of choice to investigate and characterize struc- 1:0.3
tural molecular behavior in solution. 1:0
For proteins, made of L-amino acid residues, CD is advan-
tageously used to monitor, characterize, and estimate the pro-
tein secondary structure in the far-UV region (180250 nm).
The local tertiary structure of the aromatic amino acid residues
0
observed in the near-UV region (250330 nm) can be used for
quality control as it often reveals subtle changes from batch to P3
batch often not mirrored in the far-UV region. A = (ALAR)
Another important use of CD is to probe qualitatively and 5E5
quantitatively binding interactions in a solution of biomole-
cules without the need of immobilizing and labeling them.
The application of CD spectroscopy to determine binding 10 E5
interactions is not as diffused as it should be. The main task of
this review is to facilitate and improve this diffusion.

Binding Interactions of Nonimmobilized and 300 320

Nonlabeled Biomolecules in Solution by CD Wavelength (nm)
Spectroscopy

For a hostligand (HL) complex formed from the interaction 1.5 E4

between the host (H) and the ligand (L), the apparent dissoci-
ation constant Kd1/K is determined by analyzing the CD data
A = (ALAR) (291 nm)

expressed in DA(AL
AR) using a nonlinear regression analy-

P2 (K d = 1.4 M)
sis. At equilibrium, for a single binding site, eqn [1] used to
determine the Kd has been derived from DADAHLDAHDAL, 2.0 E4
[HT][H][HL], [LT][L][HL], and Beers law Ae c l with
l1 cm:

DA DeH
DeH
KHT  KLT  1
KHT  KLT  12
4KKHT LT 
2
 2.5 E4
2K
DeH HT 
P3 (Kd = >2 mM)
[1]

Usually CD titration is conducted in a stepwise manner,

adding small aliquots of 35 ml of ligand stock solution directly
3.0 E4
into a cuvette of appropriate pathlength that contains a known
0 100 200 300
volume of the host component to be titrated. For measure-
Proline-rich peptide (M)
ments in the 190250 nm spectral region the optimum UV
absorption in a 0.05 cm pathlength is approximately 0.8, Figure 1 CD spectra of the titration of proline-rich peptides P2 (primary
whereas in the 250350 nm region the absorption in a 1 cm sequence PPPLPPKPKF) and P3 (PPPNSPRPGPPP) into aqueous GST-
pathlength is approximately 11.4. To add 1 molar equivalent SH3 protein domain (60 mM). In the 280330 nm region, both peptides

A:B
1:2.8
1:2.1
A = (ALAR)
1:1.4
1:1.1
1:0.8
1:0.4
1:0.0
260 280 300
Wavelength (nm)
A = (ALAR) (293 nm)
1:1
0.0 0.2 0.4 0.6 0.8 1.0
[p50]/{[hsp90]+[p50]}
Figure 2 CD spectra of the titration of p50 protein into hsp90 protein
(96 mM). As both proteins have plenty of aromatic residues, the CD arises
from both proteins. The 1:1 stoichiometry of the titration was revealed by
the plot DA intensity at 293 nm versus molar fraction of [p50]/{[p50]
[hsp90}.
have no CD due to the lack of Trp and Tyr aromatic residues. GST-SH3
domain shows a CD signal that is changing upon addition of P2 but not
for P3. This is unambiguously indicative of binding interactions between
GST-SH3 domain and P2 peptide. The Kd was determined by fitting the
plot DA intensity at 291 nm versus peptide concentration using a
nonlinear regression analysis from the software program Origin. In this
figure as well as in the other figures, CD spectra were measured using a
Jasco J720 spectropolarimeter purged with nitrogen gas. Multiscan
mode was used to improve the signal-to-noise ratio. This varies from
system to system but either four or nine scans with the highest scan
speed are normally used. With the J720 a scan speed of 20 nm min
1,
time constant of 4 s, and bandwidth of 2 nm were used. Using different
instruments and of different brands, it is important to find the optimum
parameters to obtain the best signal-to-noise ratio spectra. A special
1 cm cylindrical cell (Hellma, Mullheim, Germany) with a volume of
520 ml was used to carry out the CD titrations. Positive display pipettes
of 220 and 20200 ml tips (Finnpipette) were used to directly add
aliquots of ligand protein or peptide to the host protein placed in the
cuvette cell of suprasil material. The concentration was determined by
weighing using a microbalance with 1 mg sensitivity (Mettler Toledo,
Columbus, OH). The determination of accurate protein concentration is
not trivial. Special care should be taken to ensure that good accuracy
is achieved because any inaccuracy will be reflected in the determination
of the Kd as well.
Circular Dichroism, Applications 295
0.0010
A:B
1:0.0
1:0.4
1:0.8
A = (ALAR)
0.0005 1:1.1
1:1.4
1:2.1
1:2.8
260 280 300
A = (ALAR) (293 nm)
Kd = 2.5 M
0.0
0.0
[p50]
Figure 3 Difference CD spectra of the titration of p50 protein into hsp90
protein. As both proteins contribute to the CD spectrum, the Kd is
determined by subtracting from the CD spectra of hsp90 with p50 at
different molar ratios the equivalent CD contribution of p50 alone. In this
manner the plot DA versus ligand concentration reaches saturation,
otherwise it will look like the stoichiometry plot of Figure 2.
can be converted back into one active component system using
difference CD spectra. This can be obtained by subtracting
from each observed CD spectrum of the HL (1:n) mixture the
equivalent CD contribution of the ligand of n molar ratio
(Figure 3). To obtain high accuracy, it is important to terminate
the titration upon reaching saturation or as close as possible.
As a rule of thumb, a 23 molar excess of ligand is required for
stronger bindings, whereas, 46 molar excess or even higher, is
needed for weaker bindings.
Important information that can be obtained with the Kd is
the stoichiometry of the binding. This can be calculated by
plotting the intensity of differential CD or CD at single wave-
length versus ligand molar fraction (Figures 2 and 4).
The assessment of the experimental parameters, to be used
for the binding interaction study, is essential to obtain mean-
ingful and, above all, accurate results.
What is the right concentration for the host component
to be titrated by the ligand? To obtain a curve fitting like
a MichaelisMenten type of saturation curve illustrated in
Figure 1, the concentration is approximately 50100-fold that
of the dissociation constant Kd1/K. For example, a tight
binding affinity with Kd in the 1020 nM range requires a
host solution of approximately 510 mM concentration
296 Circular Dichroism, Applications

1:1.69
1:1.06 15 1:3
1:0.78
1:0.50

A = (ALAR) (105)
1:2
1:0.38
1:0.25 10 1:1
1:0.12
1:0.00 1:0.4
0 5 1:0

280 300 320

260 280 300
9

A = (ALAR) (105) = 294 nm

8
Lyz (445 M)

K d = 1.16 M
7

Lyz (17.5 M)
6

5 Kd = 1.3 M

0.0
4
Sba1 (M)
0 2 4 6

0.0 Figure 5 CD titration of a sugar ligand into lysozyme at different

1:2
concentrations. The sugar moiety is transparent in the near-UV region;
therefore, there is no need to transform the CD data into different CD
data like for Figures 2 and 3. As the affinity is concentration independent,
the Kd remains the same but the fitting is no longer of MichaelisMenten
type using the highest host concentration. With lower host
concentration the curve becomes parabolic and higher ligand excess
is required to accurately fit the binding curve.

dilution means that to measure a good signal-to-noise spec-

0.00 0.33 0.66 0.99
Figure 4 Difference CD spectra of the titration of Sba1 into preformed
hsp90AMPPNP (1:2) complex (30 mM). The 2:1 stoichiometry
indicates that one molecule of p23 protein binds to an hsp90 dimer.
whereas a medium affinity with Kd in the 110 mM range
requires a host solution of 50100 mM concentration. If the
host concentration is less than 100 Kd, the fitting curve is of
parabolic shape (Figure 5). By CD spectroscopy, the range of
the Kd of binding interactions that can be successfully deter-
mined spans from millimolar to nanomolar.
In Figure 6 are illustrated several simulated fitting curves
using Kd values from 1 nM to 10 mM. The fitted curves showing
a very sharp edge cannot be discriminated very well whereas
the more parabolic one can. The discrimination between 1 and
10 nM Kd curves (Figure 6) can be achieved by repeating the
titration with a more dilute total host concentration. The
trum a longer pathlength has to be used. Ideally the dilution
factor is used to determine the cell pathlength. In general, eqn
[1] very well fits binding interactions of one single binding site.
Failure to fit the experimental data due to wrong concentration
components is nevertheless revealing. For instance, host asso-
ciation might obscure the binding site. In this case there is a
struggle to fit the slope of the first experimental data points
reaching 1:1 molar ratio and the data can be fitted only by
reducing the total host concentration in the regression analysis.
Only the fraction of available binding site can interact with the
equivalent ligand concentration. The number of binding sites,
one or more, can be verified by an accurate determination of
the stoichiometry, which requires an accurate determination of
the host and ligand concentrations. Obviously for more than
one binding site, eqn [1] cannot be used qualitatively to deter-
mine the Kd.
Indications that more than one equilibrium might be pre-
sent in the titration can be inferred when, at high ligand molar
excesses, the experimental CD data cannot be fitted and light
scattering at longer wavelengths is observed. The host or ligand

1 nM
10 M
0.0
0.0
Ligand (M)
Figure 6 CD spectra calculated using eqn [1] of the nonlinear
regression analysis with in the software program Origin. The fitted curves
have been simulated using different values of Kd1/K to show how the
plotted curves change in terms of profile. In this example the initial total
host concentration was 10 mM. The black solid squares are the
experimental CD data point at 286 nm while the simulated CD curves have
been calculated with different Kd: 1 nM, 10 nM, 100 nM, 1 mM, 2 mM,
5 mM, and 10 mM. Note that the simulated CD spectra using 1 nM and
10 nM Kd cannot be discriminated as they superimposed one another.
8 E04 1:1
6 E04 Hsp90 + hP50
Kd = 2.5 M
4 E04
0
hP50 (M)
Figure 7 Analysis of Hsp90p50 interactions by near-UV CD. The
ligand p50 has a weaker tendency to interact with itself than to Hsp90;
hence, the titration will show two events. The first is the formation of the
Hsp90p50 complex followed by p50 association for greater 1:1 molar
excess concentration. The fitting can be still carried out and, assuming
the plateau occurred at approximately 1:1 molar ratio concentration, it
will give an apparent Kd of 2.5 mM. It is important to note that other
spectroscopic techniques will only at best give the overall picture of both
events without allowing an educated guess of the strengths of the
interactions.
components might be unstable in their unbound forms and
might precipitate once saturation is reached. The signature for
this case is the clear lack of saturation plateau (Figure 7). The
titration might be repeated in different buffer or solvent con-
ditions to solve or improve the solubility issues hence allowing
a more accurate Kd determination.
Circular Dichroism, Applications 297
Once the host concentration has been chosen, the choice of
the pathlength is dictated by the molar extinction coefficient of
the chromophore. For proteinprotein or proteinpeptide
titrations, the total concentration reached upon saturation
cannot exceed UV absorption of 0.8 in the 190260 nm (back-
bone region) and 1.21.4 in the 250340 nm spectral region
(aromatic residues and disulfide bond) otherwise signal cutoff
is reached. From Beers law, the calculation of the cell path-
length requires the knowledge of the concentration, the
absorption, and the molar extinction coefficient. For proteins
and peptides, the molar extinction coefficient can be approxi-
mately calculated from the number of Trp, Tyr, Phe amino acid
residues, and SS bonds present in the primary sequence. A set
of UV cuvette cells of different pathlength (0.001, 0.01, 0.02,
005, 0.1, 0.2, 0.5, 1, 2, and 5 cm) will allow the right choice of
pathlength, longer for dilute and shorter for concentrated solu-
tions, in order to obtain CD measurements with the highest
signal-to-noise ratio.
For small intrinsic CD signals, the choice of instrumental
parameters such as integration time or time constant, multi-
scan acquisition, increased photon flux of the light source
needs to be optimized for each titration in order to improve
the signal-to-noise ratio that is paramount to determine as
accurately as possible the binding affinity of molecular
interactions.
See also: Absolute Stereochemistry by NMR Spectroscopy;
Chiroptical Sensors; Chiroptical Spectroscopy, Emission Theory;
Chiroptical Spectroscopy, General Theory; Chiroptical Spectroscopy,
Oriented Molecules and Anisotropic Systems; Circular Dichroism and
ORD, Biomacromolecular Applications; Circular Dichroism, Induced;
Circular Dichroism Spectrometers; Circularly Polarized Luminescence
and Fluorescence Detected Circular Dichroism; Ellipsometry;
Enantiomeric Purity Studied Using NMR; Exciton Coupling; Linear
Dichroism, Applications; Linear Dichroism, Instrumentation; Magnetic
Circular Dichroism, Theory; ORD and Polarimetry Instruments; Protein
Structure Analysis by CD, FTIR, and Raman Spectroscopies; Raman
Optical Activity, Applications; Raman Optical Activity, Macromolecule
and Biological Molecule Applications; Raman Optical Activity, Small
Molecule Applications; Raman Optical Activity, Spectrometers; Raman
Optical Activity, Theory; Stereochemistry Studied Using Mass
Spectrometry; Surface-Enhanced Raman Optical Activity (SEROA); UV-
Visible Absorption and Circular Dichroism Spectroscopy, Inorganic
Chemistry Applications; Vibrational CD, Applications; Vibrational CD
Spectrometers; Vibrational CD, Theory and Application to
Determination of Absolute Configuration; Vibrational CD, Theory.
Further Reading
Barron L (2004) Molecular Light Scattering and Optical Activity, 2nd edn Cambridge
University Press.
Finley JW and Stephens PJ (1995) Density functional theory calculations of molecular
structures and harmonic vibrational frequencies using hybrid density functionals.
Journal of Molecular Structure (THEOCHEM) 357: 225235.
Freeman DJ, Patteenden G, Drake AF, and Siligardi G (1998) Marine metabolites and
metal ion chelation. Circular dichroism studies of metal binding to Lissoclinum
cyclopeptides. Journal of Chemical Society, Perkin Transactions 2: 129135.
Kim J, Huang R, Kubelka J, Bour P, and Keiderling TA (2006) Similation of infrared
spectra for b-hairpin peptides stabilized by an Aib-Gly turn sequence: Correlation
between conformational fluctuation and vibrational coupling. The Journal of
Physical Chemistry B 110: 2359023602.
298 Circular Dichroism, Applications
Nachanishi K, Berova N, and Woody RW (2000) Circular Dichroism: Principles and
Applications, 2nd edn Wiley-VCH Publisher.
Siligardi G, Hu B, Panaretou B, Piper PW, Pearl LH, and Prodromou C (2004) Co-
chaperone regulation of conformational switching in the Hsp90 ATPase cycle. The
Journal of Biological Chemistry 279: 5198951998.
Siligardi G and Hussain R (1998) Biomolecules interactions and competitions by
non-immobilised ligand interaction assay by circular dichroism. Enantiomer
3: 7787.
Siligardi G, Hussain R, Dorella-Deanna A, et al. (2002) Binding interactions of derived
HPK1 Pro-rich peptides with SH3HS1 domain suggest a possible adapter function of
HS1 protein. Proceedings of Peptides 2002, pp. 886887. Edizioni Ziino.
Siligardi G, Panaretou B, Meyer P, et al. (2002) Regulation of Hsp90 ATPase activity by
the co-chaperone Cdc37p/p50cdc37. The Journal of Biological Chemistry
23: 2015120159.