Obtaining information about protein secondary structures in

protocol
Obtaining information about protein secondary
structures in aqueous solution using Fourier
transform IR spectroscopy
Huayan Yang1, Shouning Yang1, Jilie Kong1, Aichun Dong2 & Shaoning Yu1
1Department of
Chemistry, Fudan University, Shanghai, China. 2Department of Chemistry and Biochemistry, University of Northern Colorado, Greeley, Colorado, USA.
Correspondence should be addressed to S.Y. ([email protected]) or A.D. ([email protected]).
© 2015 Nature America, Inc. All rights reserved.
Published online 5 February 2015; doi:10.1038/nprot.2015.024
Fourier transform IR (FTIR) spectroscopy is a nondestructive technique for structural characterization of proteins and polypeptides.
The IR spectral data of polymers are usually interpreted in terms of the vibrations of a structural repeat. The repeat units in
proteins give rise to nine characteristic IR absorption bands (amides A, B and I–VII). Amide I bands (1,700–1,600 cm−1) are
the most prominent and sensitive vibrational bands of the protein backbone, and they relate to protein secondary structural
components. In this protocol, we have detailed the principles that underlie the determination of protein secondary structure by
FTIR spectroscopy, as well as the basic steps involved in protein sample preparation, instrument operation, FTIR spectra collection
and spectra analysis in order to estimate protein secondary-structural components in aqueous (both H2O and deuterium oxide
(D2O)) solution using algorithms, such as second-derivative, deconvolution and curve fitting. Small amounts of high-purity (>95%)
proteins at high concentrations (>3 mg ml−1) are needed in this protocol; typically, the procedure can be completed in 1–2 d.
INTRODUCTION
Overview
IR spectroscopy is an excellent method for biological analysis1.
The introduction of computers based on the Michelson interferometer, the application of the fast Fourier transform algorithm2
and the mathematics of band-narrowing techniques (namely second-derivative analysis and Fourier self-deconvolution (FSD))
and spectral subtraction have led to major advances in IR spectroscopy, particularly in protein conformational analysis3–8.
FTIR spectroscopy measures the wavelength and intensity of the
absorption of IR radiation by a sample. The absorption of IR
radiation excites vibrational transitions in molecules. Because
the vibrational frequency and probability of absorption depend
on the strength and polarity of the vibrating bonds, they are
influenced by intramolecular and intermolecular effects.
Protein molecules exhibit many vibrational frequencies. Nine
characteristic group frequencies arise from the polypeptide
repeat unit, and they have been identified as follows: amide A
(~3,300 cm−1), amide B (~3,100 cm−1), amide I (~1,650 cm−1),
amide II (~1,550 cm−1), amide III (~1,300 cm−1), amide IV (~735
cm−1), amide V (~635 cm−1), amide VI (~600 cm−1) and amide
VII (~200 cm−1) (ref. 8). These amide vibrational bands can be
described in terms of five in-plane (C=O stretching, C-N stretching, N-H stretching, OCN bending and CNH bending) and three
out-of-plane (C-N torsion, C=O and N-H bending) displacement
coordinates9. The differential pattern in H-bonding and geometric orientations of amide bonds in α-helices, β-sheets, β-turn and
random coil structures allow the different vibration frequencies
to be associated with individual secondary structural folding10.
The amide I and amide II bands are the two major bands in the
protein IR spectrum. The most sensitive spectral region to protein
secondary structure compositions is amide I, which originates
from the C=O stretching vibration of the amide group coupled
with the in-phase bending of the N-H bond and stretching of
the C-N bond8,9.These vibrations are found between 1,700 and
382 | VOL.10 NO.3 | 2015 | nature protocols
1,600 cm−1, and they are directly related to the backbone conformation, each frequency corresponding to a particular structure. Amide II is more complex than amide I, and it is derived
mainly from in-plane N-H bending (40–60% of the potential
energy) and the C-N stretching vibration (18–40%; ref. 8).
This band is conformationally sensitive, but it has been used very
little for protein structures. The other amide bands are currently
of little use in protein structure analysis, because they are very
complex and they depend on the details of the force field, the
nature of side chains and hydrogen bonding.
FTIR is a useful tool for determining the secondary structure of proteins, and the amide I band has been widely used to
quantify the secondary structural composition of proteins and
polypeptides2,11,12. However, analysis of the spectra in terms of
protein secondary structure is not straightforward, which may
present practical problems that should be fully realized by the
practitioners. For example, amide I absorbance is usually a single
broad band; therefore, contour, band-narrowing methods4–6,13–19
and curve fitting2,18,20,21 are required to resolve the overlapping
bands. The purpose of this article is to provide a practical guide
to analyzing the secondary structure of proteins using FTIR techniques, including procedures for sample preparation, measurements, data analysis and all possible precautions that have to be
taken to ensure reproducibility. Basic background information
and information on recording and processing the FTIR spectra
is also presented. Our protocol could help researchers perform
basic experiments in this field in a manner that ensures that all
steps are performed correctly.
Comparison with other experiments.
Protein secondary structure has been investigated using a number
of spectroscopic approaches such as X-ray crystallography, NMR,
circular dichroism and mass spectrometry. Although X-ray crystallography provides detailed, atomic-level information about a
© 2015 Nature America, Inc. All rights reserved.
protocol
protein’s structure, it is not always possible
Calculate IR absorbance spectrum
to obtain crystals of high-enough quality
Dry air purge the FTIR spectroscopy
Subtract water
for such analysis. X-ray crystallography
system
also limited with respect to the extent to
Secondary-derivative spectrum or
Prepare protein: prepare protein
FSD
Collect background spectrum
which the protein environment can be
solutions for FTIR
measurement
sample
Curve fitting
modified, e.g., pH, and the differences that
Collect buffer spectrum
may exist between a protein in solution or
Check protein purity: HPLC, MS or
Peak assignment
SDS-PAGE
as a crystal. NMR offers a reasonable alternative to X-ray crystallography; however,
Data report
Determination of protein
Collect protein spectrum
concentration
at least in its present state, it is limited to
1. Prepare protein solutions for
2. Acquisition of spectra
3. Data analysis
low-molecular-weight proteins22. Circular
FTIR measurement
dichroism is one of the more common
methods used to study protein secondary Figure 1 | Schematic overview of the full protocol.
structure, but this technique has inherent
inconsistency problems in absolute secondary structure determination, and it is limited to only optically a high-quality result of protein secondary structure by FTIR is
clear solutions23–25. In comparison, FTIR has the advantage of
shown in Figure 1. The general protocol presented here has been
being a method that enables conformational analysis of a protein used successfully by our group6,37–44. Some key considerations
in aqueous solution, crystals and solids 11,26–30. Moreover, the for designing and implementing FTIR experiments for protein
high quality of FTIR spectrometers makes the absorbance at secondary structure are described in the following section.
every wavenumber valid, and the mathematical methods, such
as band-narrowing techniques and spectral subtraction, provide Key considerations in experimental design
an opportunity to obtain protein secondary structures in aqueous
Sample preparation. As water is an essential component of all
solutions11,18,26–30.
biological systems, it is important to characterize the structure
of proteins in aqueous medium. Protein samples for FTIR specApplications and limitations
troscopy must be at least 95% pure, as determined by gel electroA major advantage of FTIR spectroscopy for structural charac- phoresis (SDS–PAGE), HPLC or mass spectroscopy (MS). Protein
terization is that it is not limited by protein size or the physi- solutions need to be concentrated for successful FTIR experical state of the samples. Samples may be readily examined in ments, and this should be done after the purification steps. For
aqueous solution, hydrated films, solids, organic solvents, deter- secondary structure measurements, if water is used as the solvent,
gents, micelles and in phospholipid membranes. However, the sample concentrations should be >3 mg ml−1; meaningful results
information obtained from the analysis of solid proteins is of
can be obtained at lower concentrations (~0.5 mg ml−1) if D2O is
limited relevance owing to the nonphysiological nature of the
used. Protein concentrations can be determined by quantitative
measurement. In addition, the structure of the protein or peptide amino acid analysis, using published molar extinction co­efficients
under investigation may depend critically on the medium from if they are available, or by calculation of protein extinction
which it was dried. Standard absorption and transmission tech- coefficients from amino acid sequence data.
niques are readily applicable to protein aqueous solutions and
Using the CaF2 liquid cell. The cell used for IR spectroscopy
membrane dispersions. Many IR studies of different aspects of
should be made of a material that is transparent over the specmembrane protein structure/function or protein orientation have
tral range to be studied. Many potential window materials are
been published, and the readers can refer to other sources 31–34.
unsuitable on the basis of their water solubility (e.g., KBr and
Two types of sample accessories are typically used: windows such
NaCl) or spectral dispersion (e.g., ZnS and BaF2) (ref. 45). CaF2 is
as calcium fluoride are used for recording transmission spectra,
extensively used as a window material for IR spectroscopy owing
and an attenuated total reflectance (ATR) accessory, which is gento its very low solubility and resistance to most acids and alkalides.
erally made of zinc selenide or germanium.
There are two types of IR liquid cell. One type is a demountable
Spectral analysis using the transmission method is by far the
transmission cell, comprising rectangular windows (or circular
most popular approach for recording solution spectra of pepwindows) and an appropriate spacer (Fig. 2a). Spacers of various
tides and proteins. In transmission mode, high concentrations
thicknesses can be used to vary the cell path length. The samof proteins are required to obtain accurate protein spectra10. The
ple is introduced via filling ports (Fig. 2b). The other type of
samples are placed between windows, which are usually made
cell is formed from two rounded plates: one that is perfectly flat
from calcium fluoride. For protein adsorption studies, or for
and one with a precisely formed recess surrounded by a slightly
studies with solid powder and thin films, an ATR accessory made
deeper groove. The recessed area holds the sample against the
from zinc selenide or germanium is most commonly used for
opposing flat plate, and the surrounding groove keeps the sample
35,36
sampling
.
from becoming displaced and spreading out (Fig.2c,d; BioTools,
This protocol focuses on an approach in which FTIR with
BioCell, http://btools.com).
transmission windows is used for determining protein secondary
structure in aqueous solution. In these instances, the scope of a Recording FTIR spectra. In the transmission approach, FTIR
given protocol is generally limited to (i) peptides and proteins that spectra can be measured using an FTIR spectrometer equipped
are insoluble at high concentrations, and (ii) proteins in which the with a deuterated triglycine sulfate detector (broad band
structure changes at high concentrations. The pathway to obtain 14,000–400 cm−1). For better stability, the spectrometer should be
nature protocols | VOL.10 NO.3 | 2015 | 383
hi
on
C
us
C
aF
2
er
2
ac
Sp
C
aF
C
us
hi
on
protocol
a
c
b
d
© 2015 Nature America, Inc. All rights reserved.
Figure 2 | Photographs of CaF2 liquid cells. (a–d) Demountable transmission
cell, comprising two CaF2 windows and a spacer between them (a);
well-assembled demountable transmission cell (b); the two CaF2 windows of
the BioCell (c); and a well-assembled BioCell (d).
continuously purged with dry air. For the procedure described in
this protocol, rotameters are used to fix the flow rates as follows:
sample compartment, 5 liters per minute; analyzer compartment,
3 liters per minute.
The protein secondary structure content should be determined
from at least four independently obtained spectra, and the values
should be averaged. A sd of <0.05 when the spectrum of a protein
was correctly obtained and a full interpretation of results was
performed.
Water and water vapor. Water subtraction is a routine procedure. H2O has strong IR absorbance with three prominent bands:
~3,400, 2,125 and 1,645 cm−1. Obviously, the water absorption
band overlaps the protein vibration band, and the IR absorbance
of water is usually in the mole range, whereas protein is in the
mM or µM range. For this reason, transmission measurements
of protein solutions must be made in very thin films (<10 µm,
polyimide, Kapton); otherwise, water absorbance cannot be subtracted quantitatively. Oversubtraction or undersubtraction of the
H2O spectrum can lead to the appearance of negative and positive
bands in the final protein spectrum. This can lead to misleading
secondary structure of a peptide or protein. We therefore recommend using proteins at concentrations >3 mg ml−1 to make the
signal-to-noise (S/N) ratio of the digitized output spectra high
enough to allow water subtraction.
To avoid such artifacts, it is also essential to have an appropriate solvent blank for every sample analyzed. The blank aqueous
solvent should be analyzed under conditions identical to those
used for the protein solution. Ensuring that the cell path length,
temperature, pH, buffer concentration, number of scans and
spectral resolution are kept identical for the sample and solvent
spectrum is particularly important. Ideally, the same cell should
be used to record both the reference spectrum and the spectrum
of the protein solution under identical scan conditions; the aqueous water contribution can then be removed from the spectrum
of the protein solution using digital subtraction.
Water vapor in the sample compartment of an FTIR spectrometer can also be a problem in FTIR analysis of proteins.
Acquisition of high-quality protein spectra necessitates the
elimination of water vapor from the sample compartment of
the spectrometer, because the water vapor bands overlap with
384 | VOL.10 NO.3 | 2015 | nature protocols
the conformation-sensitive amide I band. Water vapor contributions can be reduced by purging the instrument with dry air or
nitrogen. A pre-recorded water vapor absorption spectrum can
be subtracted from the protein absorption spectrum to reduce
the vapor bands. However, subtraction of water vapor bands can
be complicated, and oversubtraction or undersubtraction can
introduce artifacts46.
To determine whether successful subtraction of absorption
bands due to liquid water and gaseous water in the atmosphere has
been achieved, a criterion is needed for determining whether the
absorption by water is correctly compensated. It has been found
that a straight baseline between 2,000 and 1,750 cm−1 can be
used as the standard to judge the success of water subtraction6,11.
When the criterion is followed and the integral are as of amide
I bands normalized, the relative areas of the second-derivative
and FSD amide I bands can be used to directly determine the
relative amounts of different types of secondary structures.
Use D2O as solvent. Traditionally, heavy water has been used in IR
spectroscopy as an alternative or complementary solvent. Unlike
H2O, D2O does not absorb in the amide I region, and it is relatively free from problems encountered with solvent subtraction.
An IR cell of large path length (up to 100 µm) can be used, which
requires much lower protein concentrations (~0.5 mg ml−1) to
obtain high-quality FTIR spectra.
Chemical bonds absorb light in the IR region according to
their dipole moment, and thus interrogation with IR spectroscopy generates a spectrum of wavenumber-associated absorbance intensities for each sample. Replacement of the hydrogen
atom in H–N and H–O bonds with a deuterium will alter the
chemical bond weights, resulting in band shifts within spectra.
Figure 3 shows the second derivative of the FTIR spectra of pyruvate kinase recorded in both H2O and D2O solution. Obviously,
there are differences between the two spectra.
A limitation of protein structural analysis in D2O is that deuterium is not a native element. The effect of H-D exchanges on protein structural properties is not fully understood, especially under
incomplete H-D exchange conditions. Furthermore, because the
exchange of D for H can affect the strength and length of hydrogen
bonds, protein secondary structures may be altered by the replacement of H2O by D2O. Therefore, H2O is much more preferable as
a solvent than D2O when studying the protein structure4,6,47,48.
Spectral processing. Three spectra are typically obtained for each
sample: a background spectrum, the spectrum for the buffer and
a spectrum for the sample itself. The spectra of buffer and protein solutions are separately corrected by obtaining a ratio with
the background spectrum; the resulting data are the absorbance
spectra for these samples. It is also important to measure the
noise in the sample spectrum. In this context, noise is defined
as the sd in a segment of the spectrum after subtraction of a
linear baseline. For noise estimation, a spectral region (preferentially without absorbance) is defined, typically 2,200–2,100 cm −1.
Averaging a large number of scans may reduce instrument noise,
so instrumental stability is important. The S/N ratio should be
>500 before spectral manipulation. In modern instruments, the
S/N ratio of the digitized output spectra is high enough to allow
the subtraction of water.
To obtain the protein spectrum, the buffer spectrum is subtracted in an iterative manner until a straight baseline is obtained
protocol
Fourier self-deconvolution (FSD). The basic theoretical consideration of FSD is that the profile of the IR vibrational band can
be a Lorentzian line shape19. The basis for deconvolution is the
transformation of a Lorentzian band of HWHH, which can be
mathematically expressed as follows:
–d2A/dv2
in H2O
in D2O
A(n ) = A0g 2 /[g 2 + (n − n 0 )2 ]
1,700
1,680
1,660
1,640
Wavenumber (nm)
1,620
1,600
© 2015 Nature America, Inc. All rights reserved.
Figure 3 | The secondary-derivative spectrum of pyruvate kinase obtained
using H2O compared with that using D2O.
in the 2,000–1,750 cm−1 spectral region. To resolve the extensively
overlapping amide I band components that arise from various
secondary structural elements, two mathematical approaches,
namely the second-derivative analysis and FSD, are often
applied13,49,50. These two methods are described in detail here.
A variety of software packages are available commercially for
these methods.
Second-derivative analysis. Derivative analysis can be used
to separate overlapping bands. The secondary derivative is
most commonly used. One of the major advantages of second­derivative analysis is that it can be performed objectively without
arbitrarily choosing deconvolutional parameters.
In the Fourier domain derivative, a spectrum A(ν) and its
Fourier transform I(x) are related by the following equation50:
∞
A(n) = F {I ( x)} = ∫ I (x)cos(2pn x)dx
0
The second-order derivative is given by:
∞
d2 A(n )/ dn 2 = ∫ (−2p x)2 I (x)cos(2pn x)dx
0
In the frequency domain derivative, the second-order derivative
of the spectrum is given by13:
AII = − (1 / pg )[2a(1 − 3an 2 )]/(1 + an 2 )3
where γ is the half-width at half-height (HWHH) and a equals
1/γ2. The peak frequency of the second derivative is identical to
the original peak frequency. The half-width of the second derivative is related to the half-width of the original line by K = γ/γII =
2.7, and the peak intensity of the second derivative is related to
the original intensity by A0II = −2 A0/γ2.
For protein secondary structure determination, the secondderivative spectrum is calculated using a seven-point Savitsky–
Golay second-derivative function. The selection of a seven-point
calculation window is due to the fact that the protein spectrum
was generally collected using 4 cm−1 resolution, which has an
~2 cm−1 per data point setting. Collectively, a seven-point
calculation window covers an ~14 cm−1 spectral region, which is
less than the half-bandwidth at half-height for almost all amide I
band components of various secondary-structural elements, and
it allows the least distortion to amide I band components.
where A0 is the maximum absorbance of the band, ν0 is the
wavenumber for A0 and γ is the HWHH. The cosine Fourier
transform of A(ν) is given as follows:
∞
I (x) = F {A(n ) = ∫ A(n )cos(2pn x)dv = 1 / 2 A0g cos(2pn 0 x) exp(−2pg x)
0
where x is a spatial frequency with units of centimeter when ν
has the units of cm−1. Because the exponential decay term exp
(−2πγx) is determined by the half-width of the vibrational band,
decreasing the rate of decay in the Fourier domain will reduce
the bandwidth in the frequency domain. This can be achieved
by multiplying I(x) with an exponentially increasing function
D(x)=exp (2πD x):
I ′(x) = I (x)exp(2pg x)Dx = 1 / 2 A0g cos(2pn 0 x)Dx
because the reverse Fourier transformation is performed on I′(x),
a narrower band A′(ν) is obtained. When using these procedures,
the resolution enhancement factor (K), which must be optimized,
is defined as the ratio
K = g /g ′
where γ is the HWHH in A(ν), and γ ′ is the HWHH in A′(ν).
Despite the apparent simplicity of FSD, this procedure presents
a number of experimental problems, which leads to frequently
encountered errors. Great care and a great deal of experience
should be exercised. The choice of values for the two key parameters (half-bandwidth and enhancement factor) is arbitrary, and
it is highly subjective because of the lack of knowledge of the
real values, and because the component bands may have unequal
half-bandwidths. Thus, by varying input parameters, substantially different FDS spectra can be generated from the same IR
spectrum49.
When FSD is used to estimate the secondary structures of
unknown samples, the initial bandwidth can only be assumed.
Underestimation of the bandwidth leads to incomplete band
separation, maintaining a high degree of Lorentzian character. Overestimation results in the appearance of undesired side
lobes. In such a situation, deconvolution will affect each component differently. Therefore, FSD is not recommended for
general use.
Quantitative analysis of the secondary structure of proteins.
The amide I band (1,700–1,600 cm−1) is highly sensitive to small
variations in molecular geometry and hydrogen bonding patterns, such that each type of secondary structure gives rise to
a different C=O stretching frequency. It is therefore especially
useful for analysis of the secondary structure composition and
conformational changes of proteins4.
nature protocols | VOL.10 NO.3 | 2015 | 385
protocol
Table 1 | Deconvoluted amide I band frequencies and assignments
to secondary structure for protein in D2O and H2O media.
© 2015 Nature America, Inc. All rights reserved.
H2Oa
D2Ob
Mean
frequencies
Assignment
Mean
frequencies
Assignment
1,624 ± 1.0
β-sheet
1,624 ± 4.0
β-sheet
1,627 ± 2.0
β-sheet
1,631 ± 3.0
β-sheet
1,633 ± 2.0
β-sheet
1,637 ± 3.0
β-sheet
1,638 ± 2.0
β-sheet
1,641 ± 2.0
310-helix
1,642 ± 1.0
β-sheet
1,645 ± 4.0
Random
1,648 ± 2.0
Random
1,653 ± 4.0
α-helix
1,656 ± 2.0
α-helix
1,663 ± 4.0
β-turn
1,663 ± 3.0
310-helix
1,671 ± 3.0
β-turn
1,667 ± 1.0
β-turn
1,675 ± 5.0
β-sheet
1,675 ± 1.0
β-turn
1,683 ± 2.0
β-turn
1,680 ± 2.0
β-turn
1,689 ± 2.0
β-turn
1,685 ± 2.0
β-turn
1,694 ± 2.0
β-turn
1,691 ± 2.0
β-sheet
1,696 ± 2.0
β-sheet
aData
are from Dong and colleagues6,47,48,54.
bData are from Susi and colleagues4,5,13,46.
The data obtained from the deconvolved and second-­derivative
spectra are used to determine the number of bands and their
positions in order to resolve the protein spectrum into its components. This is accomplished by a commercial curve-fitting
process that is commercially available from the manufacturers
of IR instruments. The fitting process is mainly controlled by
the characteristics of the individual bands, and by the position
of the baseline. The characteristics of the band are as follows:
(i) its height, (ii) its wavenumber (position of the band in the
spectrum) and (iii) the bandwidth at half-height. The program
iterates the curve-fitting process, and in each iteration the characteristic parameters (height, bandwidth, position and baseline) are
either increased or decreased in order to calculate the parameters
that will result in the best Gaussian, Lorentzian or a mixture of
Gaussian/Lorentzian-shaped curves (i.e., those that best fit the
original protein spectrum). The band area for each component
peak is used to calculate the relative contribution of component
to a particular protein secondary structure.
Quantitative analysis of the secondary structure of a protein is
based on the assumption that a protein can be considered a linear
sum of a few fundamental elements of the secondary structure.
Comparisons of the IR spectra for high-resolution X-ray crystal
structures of proteins could establish necessary spectra-­structure
correlations. Over the years, many correlations have been established between IR spectra and particular protein structures
(Table 1) (refs. 5,51).
In H2O solution, bands between 1,654 and 1,658 cm−1 are
expected for proteins with α-helical structures4,6,52. The bands
between 1,642, 1,638, 1,632, 1,627 and 1,624 cm −1 are assigned
to β-sheet components6. The bands located at 1,688, 1,680,
1,672 and 1,666 cm−1 are assigned to β-turn structures53. The
characteristic band for random coil conformation is located
at 1,648 ± 2 cm−1. For the globular protein, the correction
coefficient between IR and X-ray estimates of ordered α-helix,
ordered β-structure, β-turns and remainder was 0.98, 0.99, 0.90
and 0.92, respectively54. Dong et al.40 reported the distribution
of secondary structures determined from amide I spectra of
aqueous solutions of globular protein, which is nearly identical
to the amount computed from crystallographic data.
In D2O solution, the broad protein amide I band contours can
be decomposed into a number of components. A component centered between ~1,658 and 1,650 cm−1 has been assigned to the
α-helix8. Bands near 1,663 cm−1 are assigned to 310-helices4,8.
Bands in the 1,640–1,620 cm−1 and 1,695–1,690 cm−1 regions
have been assigned to the β-sheet by many authors4,8,7. Bands
of ~1,670, 1,683, 1,688 and 1,694 cm−1 have been proposed to
be assigned to β-turns8. Turns are also associated with a highly
characteristic band of ~1,665 cm−1. Nonordered conformation
(usually referred to as random coils) is usually associated with an
IR band of ~1,640–1,648 cm−1 (ref. 8).
Overview of the procedure
In this protocol, we describe how to set up the FTIR machines;
collect high-quality, reproducible data; and analyze the FTIR
spectra to estimate the secondary structures of proteins and
polypeptides in aqueous solution. This approach could be modified for use in other applications. The thermodynamics of protein folding can be obtained from experiments in which FTIR
data are collected as a function of temperature55. To achieve
this, the TempCon equipment (BioTools), which is connected to
the FTIR instrument, is needed. The kinetics of H-D exchange
could also be used to determine ligand-induced conformational
and structural dynamics changes56. To do this, deuteroxide D2O
should be added to lyophilized protein, and FTIR spectra of H-D
exchange should be immediately collected as a function of time.
Several reviews on FTIR spectroscopy that describe the instrumentation, theoretical background, IR data processing and analysis of data to yield secondary structure information provide
further information2,4,6,8,9,28,46,49.
MATERIALS
REAGENTS
• Ethanol, 2.5 liters (Fisher Scientific, cat. no. E/0600DF/17)
• NaCl (Sinopharm Group, CAS no. 7647-14-5)
• NH4HCO3 (Sinopharm Group, CAS no. 1066-33-7)
• CaCl2 (Sinopharm Group, CAS no. 10043-52-4)
386 | VOL.10 NO.3 | 2015 | nature protocols
• Tris-HCl (Sinopharm Group, CAS no. 5704-04-1)
• NaOH (Sinopharm Group, CAS no. 1310-73-2)
• D2O (Sinopharm Group, AS no. 7789-20-D)
• PBS (Reagent Setup)
• 5× Bradford reagent (BioLabs)
protocol
Table 2 | Instruments and corresponding data acquisition software.
Manufacturer
Instruments
Software
Website
ABB
MB3000 FTIR laboratory analyzer
BOMEM GRAMS/32
http://new.abb.com/products/measurementproducts/analytical
Thermo Fisher
Scientific
Nicolet iS10 FTIR spectrometer
OMNIC
http://www.thermoscientific.com/en/products/
fourier-transform-infrared-spectroscopy-ftir.html
Agilent Technologies
Cary 600 series FTIR spectrometers
Resolutions Pro
http://crosslab.chem.agilent.com/
© 2015 Nature America, Inc. All rights reserved.
Agilent 670-IR spectrometer
PerkinElmer
Frontier spectrometers
Spectrum Two IR spectrometers
Spectrum 10
http://www.perkinelmer.com/
JASCO UK
FTIR-4000 and FTIR-6000
FTIR spectrometers
Spectra Manager
http://www.jasco.co.uk/ftir.php
Bruker Optics
ALPHA FTIR spectrometer
OPUS
http://www.bruker.com/products.html
Shimadzu
IRTracer-100
LabSolutions IR
http://www.shimadzu.com/products/index.html
• KCl
• Na2HPO4
• KH2PO4
SDS–PAGE analysis
• Acrylamide/bisacrylamide 30% (wt/vol) solution, 29:1 (Sigma-Aldrich,
cat. no. A3574)
• SDS, 10% and 20% (wt/vol) solutions (Sigma-Aldrich, cat. nos. 71736 and
05030)
• Ammonium persulfate (Sigma-Aldrich, cat. no. A3678)
• N,N,N′,N′-tetramethylethylenediamine (TEMED; Sigma-Aldrich,
cat. no. T9281)
• Tris-HCl buffer (Sigma-Aldrich, cat. no. T3253)
• DTT or 2-mercaptoethanol (Sigma-Aldrich, cat. nos. 43817 and M6250)
• Glycerol (Sigma-Aldrich, cat. no. G5516)
• Gly (Sigma-Aldrich, cat. no. 241261)
• Coomassie Brilliant Blue R (Sigma-Aldrich, cat. no. B8647).
Example proteins (discussed in ANTICIPATED RESULTS)
• Lysozyme (from chicken egg white, lyophilized, powder; Sigma-Aldrich,
cat. no. 62970)
• Hemoglobin (from horse, lyophilized, powder; Sigma-Aldrich, cat. no. H4632)
• Myoglobin (from equine skeletal muscle, 95–100%, salt-free, lyophilized
powder; Sigma-Aldrich, M0630)
• α-Chymotrypsin (from bovine pancreas, salt-free, lyophilized powder;
Sigma-Aldrich, cat. no. C3142)
• Immunoglobulin G (from human, ≥95.0%, lyophilized powder;
Sigma-Aldrich, cat. no. 56834)
• Pyruvate kinase (from rabbit muscle, suspension in 3.2 M (NH4)2SO4
solution, pH 6; Sigma-Aldrich, cat. no. P1506)
• cAMP receptor protein (CRP; recombinant, expressed in Escherichia coli,
≥95.0%, suspension in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM
glutathione, 0.1 mM EDTA, 0.25 mM DTT, 0.1 mM PMSF, 25% (vol/vol)
glycerol, our laboratory)
• Cytochrome c (from horse heart, ≥95.0%, lyophilized powder;
Sigma-Aldrich, cat. no. C7752)
• Cu, Zn superoxide dismutase (SOD; from bovine liver, ≥95.0%, lyophilized
powder; DDI Pharmaceuticals)
EQUIPMENT
• FTIR spectroscopy equipment: for a list of commercial instruments
available, please refer to Table 2
• CaF2 liquid cells: demountable transmission cell with spacer can be
purchased from Specac (http://www.specac.com) or PerkinElmer
(http://www.perkinelmer.com/) or from any other company. Cells with two
rounded plates can be bought from BioTools under the trade name BioCell
(http://btools.com)
• Lyophilizer
• Centrifuge
• Vacuum pump
• ELISA plates, 96-well (Limbro microtitration plate)
REAGENT SETUP
Analyte proteins The proteins used as examples were chosen such that they
varied widely in terms of the relative amounts of the different possible
secondary structural features. The proteins to be analyzed should be dissolved
in appropriate buffers. Proteins with <95% purity should be purified by
HPLC or other purification technology.
Buffers Buffers are prepared according to the standard methods used
routinely in all laboratories.  CRITICAL Because ammonium salt or urea
have several strong and complicated absorption peaks in the amide I band,
and because it cannot be subtracted absolutely, any ammonium salt or urea
must be avoided. Solvent pH can be in the range of 5–10.
PBS Combine 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.8 mM
KH2PO4 (pH 7.4). This can be stored for 1 week at room temperature (20–25 °C).
EQUIPMENT SETUP
Software Spectral acquisition software is normally provided by the
instrument manufacturer. Most instrumentation software also provides a
number of pre-processing options, it often also features more advanced data
analysis options (Table 2). Various data analysis software programs and
packages are also provided by the instrument manufacturers (Table 2).
PROCEDURE
Sample preparation ● TIMING 2–10 h
1| Prepare a sample of the protein complex to be studied in buffer solution. The recommended starting amount of the
protein lyophilized powder is 300–600 µg; the total protein concentration should be in the range of 3–20 mg ml−1. For example,
use 300 µg of protein complex in 30 µl of buffer (concentration, 10 mg ml−1). The protein in buffer solution should be dialyzed
against buffer overnight to remove unwanted additives (for example, the stabilizing agent, glycerol) before being used.
nature protocols | VOL.10 NO.3 | 2015 | 387
protocol
Box 1 | Determining protein concentration using molar absorption coefficient
If the molar absorption coefficient of the protein has not been reported, calculate the molar absorption coefficient (ε) first.
The molar absorption coefficient of a protein at 280 nm, ε280 (in M−1 cm−1), is calculated using the following equation65:
e280 (mg/ml, 1 cm) = (5,500 × nTrp ) + 1, 490 × nTyr ) + 125 × ns − s
where nTrp = number of tryptophan (Trp) residues, nTyr = number of tyrosine (Tyr) residues, and nS−S = number of disulfide bonds in the
protein.
Determination of protein concentration by absorbance spectroscopy using the molar absorption coefficient
Use a UV spectrophotometer to record the absorbance of the protein solution at 280 nm (A280) and calculate the protein concentration (c)
in mg ml−1 using the following equation. The path length, l, is usually 1 cm.
© 2015 Nature America, Inc. All rights reserved.
c = (A280 × mol.wt.)/(e280 × 1)
2| Make sure that the purity of the protein is >95%. Protein purity can be determined by SDS–PAGE with some
modifications, or by HPLC or MS. Generally, SDS–PAGE, a simple and effective method, is recommended. In our hands,
SDS–PAGE gels are usually cast in two sections, a large-pore stacking gel poured on top of a small-pore separating gel,
according to the original protocol by Laemmli57.
! CAUTION Acrylamide monomer is toxic. Avoid breathing acrylamide dust, do not pipette acrylamide solutions by mouth and
wear gloves when you are handling acrylamide powder or solutions containing it. Allow unused monomer solutions to
polymerize and discard the resulting gels.
 CRITICAL STEP The optical temperature for gelation is 23–25 °C. Low temperature will result in turbid, porous and inelastic gels.
? TROUBLESHOOTING
3| Determine the concentration of the protein solutions. The following spectroscopic methods are simple, fast and reliable
for determining protein concentration. This can be done by absorbance spectroscopy using the molar absorption coefficient
(see Box 1) or by using the Bradford assay. Advice for performing the Bradford assay can, for example, be found here:
http://kitto.cm.utexas.edu/research/Kittolabpage/Protocols/Biochemistry/Bradford.html. It is worth bearing in mind that
BSA is not an ideal standard to use and that it is advisable to adjust for this by multiplying the results you get for your
protein concentration by 2.1 to get a closer approximation of your protein’s concentration.
 CRITICAL STEP Protein concentrations that are lower than that suggested will result in lower IR signal.
4| (Optional) Concentrate the protein solution if it is <3 mg ml−1 and if you want to analyze it in nondeuterated solvent.
 CRITICAL STEP To achieve the solutions necessary for successful FTIR analysis, most protein solutions need to be
concentrated; this should be carried out after the purification steps.
5| (Optional) For samples prepared in buffers in D2O, concentrations of 0.5 mg ml−1 are needed. For protein in buffer
solution in H2O, lyophilize 100 µl of protein solution at a concentration ~0.5 mg ml−1 in buffer to dryness. Reconstitute
lyophilized protein samples in 100 µl of D2O, and lyophilize the protein-D2O to dryness again. Reconstitute the protein
sample with 100 µl of D2O for the FTIR test.
 CRITICAL STEP FTIR analysis requires high-concentration protein solutions (typically >3 mg ml−1 in H2O and ~0.5 mg ml−1
in D2O) to obtain sufficient S/N ratios. Samples with protein concentrations >3 mg ml−1 in H2O and ~0.5 mg ml−1 in D2O had
equivalent FTIR spectra and secondary structures independent of the protein concentration. For proteins analyzed at much
lower concentrations, FTIR analysis may be inappropriate or, at the very least, the data will be much more difficult to
interpret quantitatively12.
Equipment preparation ● TIMING 2–8 h
6| Turn on air-conditioning and a liquid desiccant system in the room in which you are doing the experiment. Control the
temperature at 25 °C.
 CRITICAL STEP Only operate the FTIR machine in a well-enclosed room; air movement must be minimized. Maintaining a
dry and steady test environment is very important.
7| Turn on the drying equipment of the instrument.
 CRITICAL STEP Dry air or nitrogen should displace water vapor absolutely.
388 | VOL.10 NO.3 | 2015 | nature protocols
protocol
a
0.6
Single beam
8| Turn on the IR source.
 CRITICAL STEP The analyzer should be left ON at all
times. Each time the analyzer is turned OFF and then turned
back ON, you must wait for ~4 h after power-up to allow the
temperature to stabilize.
? TROUBLESHOOTING
0.2
9| Turn on the FTIR spectrometer.
? TROUBLESHOOTING
© 2015 Nature America, Inc. All rights reserved.
11| Set the data path of the operating program to store
your data.
Collecting FTIR spectra to determine secondary structures
● TIMING 1.5–3 h
12| Set the spectrometer resolution to 4 cm−1.
13| Set the spectral range from 4,000 to 500 cm−1.
14| Set the number of scans. A scan is defined as a forward
and reverse sweep of the scan mechanism. The forward and
reverse sweep information is combined during the transformation from interferograms to spectra after phase correction.
0
4,000
3,500
3,000
2,500
2,000
1,500
1,000
500
1,500
1,000
500
Wavenumber (cm–1)
b
0.8
0.6
Single beam
10| Turn on the computer and monitor, and then start the
FTIR collection program.
0.4
0.4
0.2
0
4,000
3,500
3,000
2,500
2,000
Wavenumber (cm–1)
Figure 4 | Water vapor spectrum. (a) Water vapor spectrum recorded before
purging dry air. (b) Water vapor spectrum recorded after purging the sample
compartment.
15| Select the default paths for each group of files.
16| Collect a background spectrum without IR cell. This spectrum is used to remove spectral signals that originated from air,
moisture (water vapor) and coating materials on reflecting mirrors along the IR radiation path from the spectra of protein
and buffer in order to subtract the background noise. Figure 4a shows the water vapor spectrum recorded before purging
with dry air. Figure 4b shows the spectrum recorded after purging the sample compartment.
 CRITICAL STEP If the water vapor is not purged completely by dry air, there will be peaks in the regions of 1,500–1,200 cm−1
and 4,000–3,500 cm−1. If these peaks appear, you should check the tubing and make sure that there is no air leakage;
you should check the liquid desiccant system and drying equipment, and then purge for a longer period of time.
? TROUBLESHOOTING
17| Assembling the CaF2 liquid cell. For the demountable transmission CaF2 liquid cell comprising rectangular windows
(or circular windows) and an appropriate spacer, the spacer should be placed between the two windows. Place the sample on
one window and press the second window lightly on top of the first. Spacers of various thicknesses can be used to vary the
cell path length. Typically, the path length used for protein characterization in H2O is <10 µm, usually 6 or 7.5 µm. By using
D2O, the path length can be increased to 100 µm. After the spacer is positioned appropriately, the cell can be reassembled
with four machine screws. Tighten the screws diagonally rather than in a circular direction, or tighten a single crew at a
time (Fig. 2b).
 CRITICAL STEP To obtain a high-quality IR spectrum, the spacer should be placed appropriately and smoothly.
 CRITICAL STEP The very thin spacers (6 or 7.5 µm) are extremely susceptible to electrostatic electricity. The operator
should be very careful when placing the spacer. Care should also be taken at all times when handling the cell. The CaF 2
windows are fragile. Do not stress the windows to the point of fracture. Do not use rough materials to wipe the windows;
they may cause scratches on the windows.
18| Addition of buffers or protein solutions to the cell. We describe two options for achieving this: using a demountable
transmission CaF2 liquid cell (Step 18A) or using a BioTools instrument (BioCell) (Step 18B).
 CRITICAL STEP Air bubbles and/or empty areas will result in poor absorption spectra caused by fringing from channel
spectra, dispersion of the IR beam by bubbles and/or incorrect absorbance from lack of sample. During filling, ensure that
the sample covers the entire window surface without the presence of air bubbles.
nature protocols | VOL.10 NO.3 | 2015 | 389
protocol
© 2015 Nature America, Inc. All rights reserved.
 CRITICAL STEP It is important to acquire the spectrum for the buffer before acquiring the spectrum for the sample.
The buffer used for the last dialysis step is the one most similar to that present in the protein solution, and it is
recommended to be used for background spectra.
(A) Demountable transmission CaF2 liquid cell
(i) After the demountable transmission CaF2 liquid cell is reassembled, load the cell with buffer solution using a
needleless tuberculin syringe through a loading port. The liquid solution should be slowly forced or injected into the
cell to avoid the formation of air bubbles or unfilled spaces.
(ii) Make sure that the sample fills the entire window surface without the presence of bubbles or empty cell areas.
The machine screws at the entrance and exit should be tightened to ensure that the sample remains in the cell.
(B) BioTools BioCell
(i) When you are using a BioCell, drip ~5 µl of buffer solution or protein solution onto the surface of the plate with the
recess, and then slowly cover the flat plate.
(ii) Tighten the screws diagonally rather than in a circular direction, or tighten a single screw at a time (Fig. 2d).
Care should be taken to avoid the formation of air bubbles or unfilled spaces.
! CAUTION At the end of all data acquisitions, always check that the cell did not leak. In the event of leakage from
the cell, the cell will have to either be dried or dismounted and correctly reassembled. The sample will need to be
reloaded and remeasured.
19| Collect a spectrum of the buffer solution.
20| Clean the cell. To keep the same path length, the same cell should be used to record both the reference and sample
spectra.
(A) Demountable transmission CaF2 liquid cell
(i) As for the demountable transmission CaF2 liquid cell, the cell should not be disassembled. Usually, the CaF2 cell should
be rinsed with water and ethanol and dried by suction using an aspirator or hand-operated vacuum pump.
 CRITICAL STEP Do not disassemble/reassemble the cell between the buffer and protein spectra collection, as this
will alter the path length. After collecting the spectrum of buffer, the CaF2 cell should be completely vacuum-dried to
avoid bubble formation while filling the protein sample.
(B) BioCell
(i) When you are using a BioCell, the cell can be disassembled, washed with water and ethanol and dried.
21| Collect a spectrum for the protein. Fill the cell with protein solution, as described in Step 18, allowing the solution to
be pulled through slowly. Make sure that there are no bubbles or empty cell areas.
 CRITICAL STEP The protein solution is viscous, and air bubbles can easily form. Be careful when pulling the sample into the cell.
22| Save the raw data on the hard drive or other media.
 CRITICAL STEP Always save the data immediately to prevent loss if there is a power failure.
23| Clean the cell when finished. The liquid cell may be demounted by removing the four machine screws. Wash the CaF2
window with ethyl alcohol (~5 ml volume), and then with an excess of water (~10 ml volume). Finally, wipe the windows
with lens cleaning paper gently to keep them clean.
Data analysis ● TIMING 2–4 h
24| Calculate IR absorbance spectra. Upon completion of all measurements, use the function in the software (BOMEM GRAMS/32)
to compute the absorbance spectra. After the computations are done, the results should be saved immediately.
? TROUBLESHOOTING
25| Subtract reference spectra. Subtract the reference spectra from the protein spectra to remove water contributions using a
double-subtraction procedure on the basis of the criteria described in the INTRODUCTION.
 CRITICAL STEP The flat and smooth spectral region between 2,000 and 1,750 cm−1 can be used as a criterion to judge
whether a spectrum is good. To obtain high-quality IR spectra for proteins in aqueous solution, the spectra of atmospheric
water and water in cell must be adequately subtracted from the observed protein spectrum. Any uncompensated absorption
bands from water vapor in the protein spectra will be enhanced by the derivative analysis owing to degradation of the S/N.
26| Perform baseline correction. This is a pre-processing step to account and correct for noise and sloping baseline effects.
Wavenumbers where the sample should not absorb should have an absorbance value of zero, but this is frequently not the case.
390 | VOL.10 NO.3 | 2015 | nature protocols
protocol
c
α-helix
b
Derivative
Second derivative
–d2A/dv 2
Derivative
0
–0.0002
1,700 1,690 1,680 1,670 1,660 1,650 1,640 1,630 1,620 1,610 1,600
2,000
1,900
1,800
1,700
1,600
1,500
Wavenumber (cm–1)
1,400
1,300
Wavenumber (cm–1)
1,700
1,680
β-sheet
β-turn
β-sheet
–0.0006
β-turn
–0.0004
1,660
1,640
Side chain
Original
Absorbance
a
1,620
1,600
Wavenumber (cm–1)
© 2015 Nature America, Inc. All rights reserved.
Figure 5 | IR analysis of hemoglobin. (a) IR spectra for the hemoglobin in 10 mM sodium phosphate, pH 7.4, at 20 °C. Top, the original spectrum after
subtracting liquid and gaseous water. Bottom, the secondary-derivative spectrum of the top spectrum. (b) The secondary derivative of the IR spectrum
from a. The baselines were used to obtain confined areas. (c) Curve-fitted inverted second-derivative amide I spectrum for the hemoglobin complex.
The second-derivative spectrum was inverted by factoring by −1.
This is thought to be primarily caused by Mie scattering: a type of scattering that occurs when the wavelength of IR light is
comparable or smaller than that of the biochemical structures through which it is passing58. A sloping baseline may also be
caused by sample acquisition characteristics such as temperature and concentration, or by sample reflection or instrument
anomalies. Regardless of the underlying cause, these effects can be minimized using a variety of baseline correction techniques59. Fortunately, small deviations in baseline position above or below the true baseline only subtly affect the calculated
relative percentages of major secondary structures. A popular baseline technique available with Bruker Opus software is
rubber-band baseline correction, which stretches the spectra down so that the minima in the spectral region of interest can be
used to fit a convex polygonal line (i.e., the rubber band), which is then subtracted from the original spectrum.
During this process, the small peak from 2,000 to 1,750 cm−1 resulting from the water vapor can be diminished manually.
The peak can be leveled by connecting lines through the selected points and subtracting them from the trace. This method
can remove the baseline slope and be offset by an iterative fitting process. The algorithm should be used before resolution
enhancement techniques.
27| Select the regions of interest (2,000–1,300 cm−1; Fig. 5a, original spectrum). The region of interest in the study of
protein secondary structure is 2,000–1,300 cm−1, which includes the characteristic group frequencies of amide I (~1,650 cm−1),
amide II (~1,550 cm−1) and amide III (~1,300 cm−1). A straight baseline between 2,000 and 1,750 cm−1 is a criterion for
determining whether the absorption by water is correctly subtracted.
28| Perform band narrowing of the spectrum. This can be done by performing either second-derivative analysis (Step 28A) or
Fourier deconvolution (Step 28B).
(A) Second-derivative analysis
(i) From the applications menu, select derivative function.
(ii) Use a baseline-corrected seven-point Savitsky–Golay derivative function.
(iii) Smooth and normalize amide I band area. The application of smoothing is to remove the possible white noise, which
results from the calculation by using the second derivative function. As the spectra were collected on different
instruments, even at the same protein concentration, the S/N levels were quite different, especially at lower
concentrations, the application of smoothing and normalization were necessary for second-derivative spectra.
! CAUTION Second-derivative spectra (which for ease of visual inspection are often inverted around the frequency
axis) are well suited for identifying component bands in a complex spectral region. As excessive smoothing can build
up side lobes and periodic noise, which may be confused with true spectral features, the amount of smoothing should
be kept to a minimum.
 CRITICAL STEP The key precondition of using second-derivative analysis to resolve mathematically overlapping
amide I band components of proteins is to obtain high-quality pre-deconvoluted spectra of proteins. Any absorption
peak of residual water vapor must be carefully removed with sufficient dry air purging and subsequent subtraction of
any residual water vapor by interactive subtraction of the spectrum of water vapor. The sharp and narrow absorption
peaks from residual water vapor would be greatly enhanced by the second-derivative analysis and result in distortion
of the protein amide I bands. Discard points below a threshold, and then fit the remaining points with a straight line.
nature protocols | VOL.10 NO.3 | 2015 | 391
© 2015 Nature America, Inc. All rights reserved.
protocol
(iv) Save the second-derivative spectrum.
! CAUTION Band narrowing requires the use of
unsmoothed spectra.
 CRITICAL STEP The smoothness and featurelessness
of the nonstructural region above 1,750 cm−1 in the
second-derivative spectra can be used as an indicator
of successful water subtraction (Fig. 5a,
secondary-derivative spectrum)49,60.
(B) FSD
(i) Choose FSD from the Process menu.
(ii) Set the parameter for half-bandwidth and enhancement
factor. Usually, you can set a half-bandwidth first,
and then change the enhancement factor; if you are
1,700
1,680
1,660
1,640
1,620
1,600
not satisfied with the result, you can change the
Wavelength (nm)
enhancement factor from 1.5 to 3.5. If none of the
Figure 6 | An example of a poorly curve-fitted spectrum.
results are satisfactory, you can try another halfbandwidth and set the enhancement factor again.
Usually, varying half-bandwidth in the range of
15–30 cm−1 and varying enhancement factors, commonly referred to as K values, in the range of 1.5–3.5 results in
proper spectral band resolution.
(iii) When you are satisfied with the resulting spectrum, replace the original spectrum with it or add it to a
spectral window.
(iv) Save the FSD spectrum and record the half-bandwidths and enhancement factor.
 CRITICAL STEP If negative peaks appear in the result, it means that they have been over-resolved. You can adjust
the bandwidth and enhancement factor to eliminate this problem.
29| (Optional) After second-derivative calculation or FSD calculation, truncate the second-derivative spectrum to 1,700
and 1,600 cm−1 sections (extended amide I region) by selecting the Utilities function from the applications menu and then
selecting the ZAP function.
30| Nonzero baseline. A nonzero baseline can be easily obtained by connecting the two (up to four, in proteins containing
>60% α-helical structure, such as hemoglobin and myoglobin) most positive points within the amide I region (Fig. 5b).
31| Perform a baseline correction and set the nonzero baseline as ZERO for further analysis, such as curve fitting.
32| Peak fitting. Carry out peak fitting using nonlinear least squares61. The frequencies of the band centers found in the
second-derivative spectra in the amide I regions can be used as starting points for Gaussian curve fitting. In most instances,
we have found that the discrepancies between component frequencies obtained by second derivatization and Gaussian curve
fitting are <2 cm−1 (Fig. 5c). The assignment of peaks is based on previous work (Table 1). Over the years, many correlations
have been established between IR spectra and particular protein structures. Both amino acid side chains and experimental
procedures may make contributions to the intensity of characteristic protein amide bands; therefore, caution has to be
exercised when interpreting the IR spectra of proteins.
 CRITICAL STEP Some of the curves available for peak fitting are asymmetrical. If none of these exactly fit your peak, then
you will need to use the peak shape that gives the best match.
 CRITICAL STEP In extreme cases, the fit may even ‘walk away’ to a ridiculous solution (Fig. 6 shows an example). The only
way to recover from this problem is to re-enter the starting fit values.
? TROUBLESHOOTING
33| Save the peak-fitting results. You can save the peak-fitting results as an Excel document after performing the peak fitting.
The report content can include the following: Area above baseline—reports each baseline corrected; Center X—reports peak
center position (peak center frequency); Absolute peak height from the Y = 0 level; and Percentage area of each peak relative
to the sum area of all peaks within that group.
34| Quantification of protein secondary structures. Calculate the secondary content from the areas of the individual assigned
bands and their fraction of the total area in the amide I regions. In the second-derivative spectrum, the peak frequency
of an absorbance is identical to the original peak frequency, and the areas correspond to the different types of secondary
392 | VOL.10 NO.3 | 2015 | nature protocols
protocol
s­ tructures. First, sum the total peak areas and then calculate each peak area as a percentage of the sum of all peak areas.
The secondary structure content should be determined from at least four independently obtained spectra, and the values
should be averaged.
? TROUBLESHOOTING
35| Save the results calculated for the secondary structure. Save the final calculated results.
? TROUBLESHOOTING
Troubleshooting advice can be found in Table 3.
© 2015 Nature America, Inc. All rights reserved.
Table 3 | Troubleshooting table.
Step
Problem
Solution
2
Insufficient sample purity
Improve the complex isolation/purification procedure
8
The lamp does not turn on
In some instruments, the lamp base needs to warm up before firing. Wait for a minute and try
again. If this does not work, check the manual; you might need a service call
9
The machine does not turn on
First, make sure that everything is plugged in and that the computer and monitor are turned
on. Consult the manual to see how to start the collection program if it does not start
automatically when the machine is turned on
16
There is a burr peak in the
region of 1,500–1,200 cm−1 and
4,000–3,500 cm−1
Check the pipe and make sure that there is no air leakage, and then continue to purge for a
longer period of time
24
There is a very low-IR
absorbance signal or no signal
Check the protein concentration and quality using SDS–PAGE staining or Bradford assay.
Use larger amounts of protein or try to increase the protein concentration
Make sure that there are no bubbles or empty cell areas when collecting the sample spectrum
32
When initial peaks are estimated This could be because of the internal baseline estimator. When the peaks are initially set,
in peak fit, assigned peaks go
the baseline is estimated by a straight line connecting the end points of the spectrum in the
below the baseline
displayed region. If one side of the region is substantially higher than the other, the baseline
could go diagonally right through the middle of the peak data. The peak height values are initially calculated as the Y delta from this estimated baseline to the set position for the top of
the peak. If this is below the estimated baseline, it will make a ‘negative’ peak
There are a couple of solutions. First, if the baseline slope in the fitting region is really that
bad, then the baseline should be processed first
Alternatively, turn off the baseline fitting option and do the fit without a baseline. However,
be aware that the calculations of the peaks will be in serious error if there is any offset of tilt
to the data baseline
Finally, make sure that the edges of the fitting region are only baseline data and not part of
another peak. Remember that the baseline is initially estimated by connecting the left and
right data points in the fitting region with a straight line
34
During the iteration in peak
fitting, get a ‘Not Converged’
message
The program uses the Levenberg-Marquardt algorithm to perform the curve fit by attempting to fit
equations to the line in the data (i.e., the curve fit). The algorithm may find many different equations that match the data, and it may try to minimize the difference between the calculated line
and the real data line. The final one that is saved is the one that gives the smallest difference
It is possible for the program to stop at a minimum that is not the best fit, as determined by
the parameters required for a convergence. If you find that the program is stuck at a minimum
and will not give you a better fit, you can, if the fit looks good, accept the fit as is. If the
fit is bad, a possible solution is to go back into the Peak Editor and manually move a peak
to a different position and then return to the Fit section and run it again. If there is still no
improvement, then continue to change other parameters such as peak shape or position until
either an ‘acceptable minimum’ is found or the fit has converged
There is a big variability (>5%)
between the four obtained
independently acquired spectra
Clean the cell very gently. Although the same cell was used to record the spectra of
background and protein samples and the CaF2 cell was cleaned using a vacuum pump,
the pathlength will be altered slightly
Any variation in the accuracy of subtracting the water band should be taken into account
nature protocols | VOL.10 NO.3 | 2015 | 393
protocol
● TIMING
Summary of the procedure for collecting and analyzing one FTIR spectrum, which can be completed in ~1–2 d
Steps 1–5, protein sample preparation: 2–10 h (can be done in advance)
Steps 6–11, machine setup: 2–8 h (the amount of time depends on air humidity and drying equipment of the laboratory)
Steps 12–23, data collection (assuming that three replicate data sets are collected for protein and buffer solutions): 1.5–3 h
Steps 24–35, data analysis (calculation of IR absorbance spectra; subtraction of buffer spectra; baseline correction;
obtaining second-derivative spectrum; obtaining FSD spectrum; curve-fitting process; and giving resulted report): 2–4 h
ANTICIPATED RESULTS
On the basis of our experience, the second-derivative analysis, FSD and curve-fitting methods used to analyze protein spectra
should give reasonable determinations of the secondary structure contents of proteins (Table 4). Because no experimental
method is currently available that permits an absolutely accurate determination of the secondary structure of proteins in
solution, the accuracy of the amounts we estimate to be present based on X-ray crystallography.
© 2015 Nature America, Inc. All rights reserved.
The -helix protein myoglobin
Myoglobins (Mbs) are monomeric molecules that contain a single-chain polypeptide. It has been known for decades from
X-ray crystallographic studies that Mbs contain 75–85% α-helical structure49. The IR absorbance spectrum of Mbs exhibits its
amide I and amide II band maxima at 1,654 and 1,548 cm−1, respectively. Quantitative analysis was carried out with both refined second-derivative/curve-fitting method (Fig. 7a) and the FSD/curve-fitting analysis method. Both methods give similar
results (~80% α-helix). However, the secondary derivative has overwhelming advantages: it separates amide I components
without any arbitrary factor input, such as half-bandwidths and enhancement factor.
The -sheet and loop structures of Cu, Zn SOD
The Cu, Zn SOD is a dimeric enzyme that is composed of two noncovalently bound identical subunits; each subunit contains
one copper and one zinc ion. X-ray crystal structures are known for the oxidized enzyme. Each subunit is composed primarily
of eight antiparallel β-strands that form a flattened β-barrel plus three external loops. The IR absorbance spectrum of
oxidized Cu, Zn SOD exhibited an amide I band maximum at 1,648 cm−1 and a strong shoulder near 1,634 cm−1. By secondary
derivative or FSD separation of the overlapping amide I, three bands (1,694, 1,633 and 1,625 cm−1) can be ascribed to
Table 4 | Percentage distributions of secondary structures of nine proteins determined by FTIR spectroscopy and X-ray data.
Secondary structures (%)
a-helix
Protein
b-sheet
Fourier
selfSecond deconvoX-raya derivative lution X-raya
Second
derivative
Random
b-turn
Fourier
Fourier
Fourier
selfselfSecond
selfdeconvoSecond deconvoderiva- deconvolution X-raya derivative lution X-raya tive
lution
Myoglobin
85
82
85
0
12
7
8
5
8
7
0
0
Hemoglobin
87
85
78
0
6
12
7
9
10
8
0
0
Lysozyme
45
52
40
19
12
19
23
19
27
13
17
14
Pyruvate kinase
41
43
40
33
52
37
14
5
13
12
0
10
Cytochrome c
48
51
42
10
18
21
17
14
25
25
9
12
3
7
3
67
65
64
18
13
28
12
11
5
37
41
35
38
36
42
19
20
16
6
3
6
Cu, Zn SOD
4
13
16
34
35
32
21
21
20
41
31
32
β-Lactoglobulin A
7
9
11
53
54
52
—
27
27
—
9
10
23
19
15
46
49
40
21
16
36
—
—
9
Immunoglobulin G
CRP
RNase A
aX-ray
data are obtained from Levitt and Greer64 and FTIR data are obtained from our laboratory.
394 | VOL.10 NO.3 | 2015 | nature protocols
protocol
1,660
1,620
Wavenumber (cm–1)
© 2015 Nature America, Inc. All rights reserved.
1,600 1,700
1,680
1,660
1,640
Wavenumber (cm–1)
1,620
1,600
1,700
β-sheet
1,680
Other
1,660
1,640
1,620
Other
α-helix
Other
β-turn
β-turn
β-turn
β-sheet
β-sheet
FSD
β-turn
β-turn
Other
β-sheet
β-sheet
1,640
β-sheet
Random
FSD
α-helix
1,680
310-helix
β-turn
1,700
β-sheet
α-helix
β-turn
Other
β-sheet
β-sheet
α-helix
β-turn
β-turn
FSD
SD
β-sheet
Random
α-helix
SD
SD
α-helix
c
b
β-turn
a
1,600
Wavenumber (cm–1)
Figure 7 | Second-derivative and FSD IR spectrum for three example proteins. (a) Myoglobins in PBS buffer solution. Top, the inverted second-derivative (SD)
spectrum. Bottom, the FSD spectrum. The curve fitting was carried out using the BOMEM GRAMS/32 software. (b) Cu, Zn superoxide dismutase in Tris buffer
solution. Top, the inverted SD spectrum. Bottom, The FSD spectrum. The curve fitting was carried out using the BOMEM GRAMS/32 software. Adapted from
ref. 62 with permission from Elsevier. (c) CRP in Tris buffer solution. Top, the inverted SD spectrum. Bottom, the FSD spectrum. The curve fitting was carried
out using the BOMEM GRAMS/32 software. Adapted with permission from ref. 43, © 2002, American Chemical Society.
β-sheet structure and three bands (1,686, 1,677 and 1,669 cm−1) to β-turn structure. The bands located at 1,648 ± 2.0 cm−1
can be assigned to the unordered structure. Quantitative analysis of these spectra indicates that oxidized (CuII, ZnII) enzymes
are composed of ~35% antiparallel β-sheet, 45% unordered/loop and 20% β-turn structures (Fig. 7b)62.
The case of / structures of CRP
CRP is a dimeric protein that is composed of two chemically identical subunits; each subunit consists of a larger N-terminal
domain in which resides the cAMP-binding site and a smaller C-terminal domain, which binds to DNA through a helixturn-helix motif. X-ray crystallographic studies revealed that CRP contains 37% α-helical, 38% β-sheet, 19% β-turn and 6%
unordered/loop, respectively43,63. The IR absorbance spectrum of CRP exhibited an amide I band maximum at 1,650 cm−1.
By secondary derivative or FSD separation, the overlapping amide I band at 1,656 cm−1 can be ascribed to α-helical structure;
the band at 1,637 cm−1 can be ascribed to β-sheet structure; the bands at 1,685 and 1,673 cm−1 can be ascribed to β-turn
structure; and the band at 1,646 cm−1 can be ascribed to unordered structure. Quantitative analysis of these spectra indicates
that CRPs are composed of ~41% α-helical, 36% β-sheet, 20% β-turn and 3% unordered/loop, respectively (Fig. 7c).
Acknowledgments This project was supported in part by grants from
the National Natural Science Foundation of China (nos. 21275032, 31470786
and 21335002).
AUTHOR CONTRIBUTIONS H.Y. and S. Yu designed, implemented and wrote
the protocol. S. Yang and J.K. implemented the protocol. A.D. contributed to
data analysis.
COMPETING FINANCIAL INTERESTS The authors declare no competing financial
interests.
Reprints and permissions information is available online at http://www.nature.
com/reprints/index.html.
1. Baker, M.J. et al. Using Fourier transform IR spectroscopy to analyze
biological materials. Nat. Protoc. 9, 1771–1791 (2014).
2. Arrondo, J.L., Muga, A., Castresana, J. & Goni, F.M. Quantitative studies
of the structure of proteins in solution by Fourier-transform infrared
spectroscopy. Prog. Biophys. Mol. Biol. 59, 23–56 (1993).
3. Purcell, J.M. & Susi, H. Solvent denaturation of proteins as observed by
resolution-enhanced Fourier transform infrared spectroscopy. J. Biochem.
Biophys. Methods 9, 193–199 (1984).
4. Susi, H. & Byler, D.M. Resolution-enhanced Fourier transform infrared
spectroscopy of enzymes. Methods Enzymol. 130, 290–311 (1986).
5. Byler, D.M. & Susi, H. Examination of the secondary structure of proteins
by deconvolved FTIR spectra. Biopolymers 25, 469–487 (1986).
6. Dong, A., Huang, P. & Caughey, W.S. Protein secondary structures in
water from second-derivative amide I infrared spectra. Biochemistry 29,
3303–3308 (1990).
7. Lee, D.C., Haris, P.I., Chapman, D. & Mitchell, R.C. Determination of
protein secondary structure using factor analysis of infrared spectra.
Biochemistry 29, 9185–9193 (1990).
8. Krimm, S. & Bandekar, J. Vibrational spectroscopy and conformation of
peptides, polypeptides, and proteins. Adv. Protein Chem. 38, 181–364
(1986).
9. Bandekar, J. Amide modes and protein conformation. Biochim. Biophys.
Acta 1120, 123–143 (1992).
10. Singh Bal, R. in Infrared Analysis of Peptides and Proteins Vol. 750 2–37
(American Chemical Society, 1999).
11. Kong, J. & Yu, S. Fourier transform infrared spectroscopic analysis of
protein secondary structures. Acta Biochim. Biophys. Sin. (Shanghai) 39,
549–559 (2007).
12. Jiang, Y. et al. Qualification of FTIR spectroscopic method for
protein secondary structural analysis. J. Pharm. Sci. 100, 4631–4641
(2011).
13. Susi, H. & Michael Byler, D. Protein structure by Fourier transform infrared
spectroscopy: second derivative spectra. Biochem. Biophys. Res. Commun.
115, 391–397 (1983).
nature protocols | VOL.10 NO.3 | 2015 | 395
© 2015 Nature America, Inc. All rights reserved.
protocol
14. Lee, D.C., Hayward, J.A., Restall, C.J. & Chapman, D. Second-derivative
infrared spectroscopic studies of the secondary structures of
bacteriorhodopsin and Ca2+-ATPase. Biochemistry 24, 4364–4373 (1985).
15. Yang, W.-J., Griffiths, P.R., Byler, D.M. & Susi, H. Protein conformation
by infrared spectroscopy: resolution enhancement by Fourier
self-deconvolution. Appl. Spectrosc. 39, 282–287 (1985).
16. Olinger, J.M., Hill, D.M., Jakobsen, R.J. & Brody, R.S. Fourier transform
infrared studies of ribonuclease in H2O and 2H2O solutions. Biochim.
Biophys Acta 869, 89–98 (1986).
17. Surewicz, W.K., Mantsch, H.H., Stahl, G.L. & Epand, R.M. Infrared
spectroscopic evidence of conformational transitions of an atrial
natriuretic peptide. Proc. Natl. Acad. Sci. USA 84, 7028–7030 (1987).
18. Griebenow, K. & Klibanov, A.M. On protein denaturation in aqueousorganic
mixtures but not in pure organic solvents. J. Am. Chem. Soc. 118,
11695–11700 (1996).
19. Kauppinen, J.K., Moffatt, D.J., Mantsch, H.H. & Cameron, D.G. Fourier
self-deconvolution: a method for resolving intrinsically overlapped bands.
Appl. Spectrosc. 35, 271–276 (1981).
20. Ruegg, M., Metzger, V. & Susi, H. Computer analyses of characteristic
infrared bands of globular proteins. Biopolymers 14, 1465–1471 (1975).
21. Fafarman, A.T. et al. Thiocyanate-capped nanocrystal colloids: vibrational
reporter of surface chemistry and solution-based route to enhanced coupling
in nanocrystal solids. J. Am. Chem. Soc. 133, 15753–15761 (2011).
22. Lynch, I., Dawson, K.A. & Linse, S. Detecting cryptic epitopes created by
nanoparticles. Sci. Signal. 2006, pe14 (2006).
23. van Stokkum, I.H., Spoelder, H.J., Bloemendal, M., van Grondelle, R. &
Groen, F.C. Estimation of protein secondary structure and error analysis
from circular dichroism spectra. Anal. Biochem. 191, 110–118 (1990).
24. Matsuo, K., Yonehara, R. & Gekko, K. Improved estimation of the
secondary structures of proteins by vacuum-ultraviolet circular dichroism
spectroscopy. J. Biochem. 138, 79–88 (2005).
25. Greenfield, N.J. Using circular dichroism spectra to estimate protein
secondary structure. Nat. Protoc. 1, 2876–2890 (2006).
26. Arrondo, J.L.R. & Goñi, F.M. Structure and dynamics of membrane proteins
as studied by infrared spectroscopy. Prog. Biophys. Mol. Biol. 72, 367–405
(1999).
27. Goormaghtigh, E., Raussens, V. & Ruysschaert, J.M. Attenuated total
reflection infrared spectroscopy of proteins and lipids in biological
membranes. Biochim. Biophys. Acta 1422, 105–185 (1999).
28. Barth, A. Infrared spectroscopy of proteins. Biochim. Biophys. Acta 1767,
1073–1101 (2007).
29. Liu, K.Z., Shaw, R.A., Man, A., Dembinski, T.C. & Mantsch, H.H. Reagentfree, simultaneous determination of serum cholesterol in HDL and LDL by
infrared spectroscopy. Clin. Chem. 48, 499–506 (2002).
30. Lenk, T.J., Horbett, T.A., Ratner, B.D. & Chittur, K.K. Infrared
spectroscopic studies of time-dependent changes in fibrinogen adsorbed
to polyurethanes. Langmuir 7, 1755–1764 (1991).
31. Goormaghtigh, E., Cabiaux, V. & Ruysschaert, J.M. Determination of soluble
and membrane protein structure by Fourier transform infrared spectroscopy.
III. Secondary structures. Subcell. Biochem. 23, 405–450 (1994).
32. Tamm, L.K. & Tatulian, S.A. Infrared spectroscopy of proteins and peptides
in lipid bilayers. Q. Rev. Biophys. 30, 365–429 (1997).
33. Arrondo, J.L. & Goni, F.M. Structure and dynamics of membrane proteins
as studied by infrared spectroscopy. Prog. Biophys. Mol. Biol. 72, 367–405
(1999).
34. Chapman, D., Jackson, M. & Haris, P.I. Investigation of membrane protein
structure using Fourier transform infrared spectroscopy. Biochem. Soc.
Trans. 17, 617–619 (1989).
35. Angeletti, R.H. (ed.) Techniques in Protein Chemistry III. (Academic
Press, 1992).
36. Chittur, K.K. FTIR/ATR for protein adsorption to biomaterial surfaces.
Biomaterials 19, 357–369 (1998).
37. Yu, S. et al. Solution structure and structural dynamics of envelope
protein domain III of mosquito-and tick-borne flaviviruses. Biochemistry
43, 9168–9176 (2004).
38. Shen, X. et al. The secondary structure of calcineurin regulatory region
and conformational change induced by calcium/calmodulin binding.
J. Biol. Chem. 283, 11407–11413 (2008).
39. Yu, S., Mei, F.C., Lee, J.C. & Cheng, X. Probing cAMP-dependent protein
kinase holoenzyme complexes Iα and IIβ by FT-IR and chemical protein
footprinting. Biochemistry 43, 1908–1920 (2004).
40. Dong, A., Huang, P. & Caughey, W.S. Redox-dependent changes in
β-extended chain and turn structures of cytochrome c in water solution
determined by second derivative amide I infrared spectra. Biochemistry
31, 182–189 (1992).
396 | VOL.10 NO.3 | 2015 | nature protocols
41. Dong, A. et al. Infrared and circular dichroism spectroscopic
characterization of structural differences between β-lactoglobulin A and B.
Biochemistry 35, 1450–1457 (1996).
42. Dong, A., Matsuura, J., Manning, M.C. & Carpenter, J.F. Intermolecular
β-sheet results from trifluoroethanol-induced nonnative α-helical
structure in β-sheet predominant proteins: Infrared and circular dichroism
spectroscopic study. Arch. Biochem. Biophys. 355, 275–281 (1998).
43. Dong, A., Malecki, J.M., Lee, L., Carpenter, J.F. & Lee, J.C. Ligand-induced
conformational and structural dynamics changes in Escherichia coli cyclic
AMP receptor protein. Biochemistry 41, 6660–6667 (2002).
44. Dong, A. et al. Secondary structure of recombinant human cystathionine
β-synthase in aqueous solution: Effect of ligand binding and proteolytic
truncation. Arch. Biochem. Biophys. 344, 125–132 (1997).
45. Nasse, M.J., Ratti, S., Giordano, M. & Hirschmugl, C.J. Demountable
liquid/flow cell for in vivo infrared microspectroscopy of biological
specimens. Appl. Spectrosc. 63, 1181–1186 (2009).
46. Surewicz, W.K. & Mantsch, H.H. New insight into protein secondary
structure from resolution-enhanced infrared spectra. Biochim. Biophys.
Acta 952, 115–130 (1988).
47. Kalnin, N.N., Baikalov, I.A. & Venyaminov, S. Quantitative IR
spectrophotometry of peptide compounds in water (H2O) solutions. III.
Estimation of the protein secondary structure. Biopolymers 30, 1273–1280
(1990).
48. Venyaminov, S. & Kalnin, N.N. Quantitative IR spectrophotometry of
peptide compounds in water (H2O) solutions. II. Amide absorption bands
of polypeptides and fibrous proteins in α-, β-, and random coil
conformations. Biopolymers 30, 1259–1271 (1990).
49. Dong, A., Huang, P., Caughey, B. & Caughey, W.S. Infrared analysis of
ligand- and oxidation-induced conformational changes in hemoglobins and
myoglobins. Arch. Biochem. Biophys. 316, 893–898 (1995).
50. Cameron, D.G. & Moffatt, D.J. A generalized approach to derivative
spectroscopy. Appl. Spectrosc. 41, 539–544 (1987).
51. Sarver, R.W. Jr & Krueger, W.C. Protein secondary structure from Fourier
transform infrared spectroscopy: a database analysis. Anal. Biochem. 194,
89–100 (1991).
52. Holloway, P.W. & Mantsch, H.H. Structure of cytochrome b5 in solution
by Fourier-transform infrared spectroscopy. Biochemistry 28, 931–935
(1989).
53. Chou, P.Y. & Fasman, G.D. β-turns in proteins. J. Mol. Biol. 115, 135–175
(1977).
54. Venyaminov, S. & Kalnin, N.N. Quantitative IR spectrophotometry of
peptide compounds in water (H2O) solutions. I. Spectral parameters
of amino acid residue absorption bands. Biopolymers 30, 1243–1257
(1990).
55. Dong, A., Randolph, T.W. & Carpenter, J.F. Entrapping intermediates of
thermal aggregation in α-helical proteins with low concentration of
guanidine hydrochloride. J. Biol. Chem. 275, 27689–27693 (2000).
56. Yu, S., Fan, F., Flores, S.C., Mei, F. & Cheng, X. Dissecting the mechanism
of Epac activation via hydrogen–deuterium exchange FT-IR and structural
modeling. Biochemistry 45, 15318–15326 (2006).
57. Laemmli, U.K. Cleavage of structural proteins during the assembly of the
head of bacteriophage T4. Nature 227, 680–685 (1970).
58. Bassan, P. et al. Resonant Mie scattering in infrared spectroscopy of
biological materials: understanding the ‘dispersion artefact’. Analyst 134,
1586–1593 (2009).
59. Kelly, J.G. et al. Biospectroscopy to metabolically profile biomolecular
structure: a multistage approach linking computational analysis with
biomarkers. J. Proteome Res. 10, 1437–1448 (2011).
60. Prestrelski, S.J., Byler, D.M. & Liebman, M.N. Comparison of various
molecular forms of bovine trypsin: correlation of infrared spectra with
X-ray crystal structures. Biochemistry 30, 133–143 (1991).
61. Savitzky, A. & Golay, M.J.E. Smoothing and differentiation of data
by simplified least squares procedures. Anal. Chem. 36, 1627–1639
(1964).
62. Dong, A., Huang, P. & Caughey, W.S. Redox-dependent changes in β-sheet
and loop structures of Cu,Zn superoxide dismutase in solution observed by
infrared spectroscopy. Arch. Biochem. Biophys. 320, 59–64 (1995).
63. Sharma, H., Yu, S., Kong, J., Wang, J. & Steitz, T.A. Structure of apo-CAP
reveals that large conformational changes are necessary for DNA binding.
Proc. Natl. Acad. Sci. USA 106, 16604–16609 (2009).
64. Levitt, M. & Greer, J. Automatic identification of secondary structure in
globular proteins. J. Mol. Biol. 114, 181–239 (1977).
65. Pace, C.N., Vajdos, F., Fee, L., Grimsley, G. & Gray, T. How to measure and
predict the molar absorption coefficient of a protein. Protein Sci. 4,
2411–2423 (1995).