Structure-energy relations in hen egg white lysozyme observed during refolding from a quenched unfolded state

Article abstract of DOI:10.1039/b907656e

We use infrared spectroscopy to study the evolution of protein folding intermediate structures on arbitrarily slow time scales by rapidly quenching thermally unfolded hen egg white lysozyme in a glassy matrix, followed by reheating of the protein to refold; upon comparison with differential scanning calorimetric experiments, low-temperature structural changes that precede the formation of energetic native contacts are revealed. The Royal Society of Chemistry 2009.

Full text of DOI:10.1039/b907656e

View Article Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/chemcomm | ChemComm
Structure–energy relations in hen egg white lysozyme observed during
refolding from a quenched unfolded statew
Theresa Y. Cho,a Nolene Byrne,b David J. Moore,c Brian A. Pethica,a C. Austen Angellb
and Pablo G. Debenedetti*a
Received (in Cambridge, UK) 16th April 2009, Accepted 18th May 2009
First published as an Advance Article on the web 12th June 2009
DOI: 10.1039/b907656e
We use infrared spectroscopy to study the evolution of protein
folding intermediate structures on arbitrarily slow time scales by
rapidly quenching thermally unfolded hen egg white lysozyme in
a glassy matrix, followed by reheating of the protein to
refold; upon comparison with differential scanning calorimetric
experiments, low-temperature structural changes that precede
the formation of energetic native contacts are revealed.
(IR) is used to collect spectra of the protein at various
temperatures. Intermediates are observed and captured by
taking advantage of the very slow kinetics in the viscous liquid
state near the glass temperature (Tg).6
Angell and Wang first introduced the quench-and-refold
approach and provided proof-of-concept using differential
scanning calorimetry (DSC).5 In conventional DSC studies,
a protein is considered fully denatured upon the completion of
energy absorption, as indicated by a peak centered at the
protein denaturation temperature (Td). Refolding of a protein
is also indicated by the energy released, calculated as the area
under the peak.7,8 If unfolding and refolding energies are the
same, the protein is said to have refolded completely. Combin-
ing the quench-and-refold approach with DSC, detection of
stages during refolding is possible.5 Here, we show that
combining DSC and IR offers new insight into energetic
and structural signatures of the refolding process, and their
non-trivial relationship to each other.
The folding of proteins into their functional state is one of
nature’s most ubiquitous phenomena. Understanding the
mechanisms underlying protein folding is important for the
rational modification and design of novel proteins, the under-
standing of human degenerative diseases that are tied to
protein misfolding and/or aggregation,1,2 and the development
of improved formulations to prevent protein degradation
during storage.
Current optical methods for studying folding mechanisms
with sub-ms resolution do not probe folding from a fully
denatured state.3,4 Standard stopped-flow techniques, on the
other hand, are able to prompt folding from fully-denatured
samples, but have a time resolution of the order of 1 ms.3,4
Consequently, important events in the refolding of fully
denatured states that occur on the ms time scale are not
observable, which creates an important gap in fundamental
understanding. Recent developments in microfluidics have
allowed denaturant dilution in less than 20 ms, offering the
prospect of vastly improved time resolution using stopped-
flow approaches.3,4 Here we present an experimental method
that is capable of detecting the changes in intermediate
structure during the folding process on arbitrarily slow (arrested)
time scales, and starting from the fully thermally-denatured state.
To this end, a recently introduced quench-and-refold approach5
was adapted to spectroscopic scrutiny. The approach consists
of the following steps: (1) thermally unfold/reversibly denature
the protein, (2) trap the unfolded protein into a glassy matrix
by quenching the sample in liquid nitrogen (LN2), and (3)
refold the protein by slowly heating the sample across the glass
transition temperature. At this stage, infrared spectroscopy
The model protein must have the ability to refold com-
pletely and be a slow folder (4100 ms), allowing immersion in
LN2 with a 120 K/s quenching rate for small samples9 to
suppress folding during quenching so as to trap the unfolded
protein into a glassy state. Electrospray is one possible tech-
nique to quench fast-folding proteins.9 Hen egg white
lysozyme was chosen as the model protein, as it is commonly
used in experiments and is known to exhibit intermediates.10,11
The solvent must not allow ice to form upon cooling or
heating, because this would cause abrupt changes in solution
composition and sample temperature. The solvent itself should
not denature the proteins at room temperature and should
support the thermally denatured protein at higher tempera-
tures without permitting aggregation to occur. The solvent’s
melting and glass transition temperatures (Tm, Tg) are also
important in experimental design. The sample solution must
be homogenous at the protein unfolding temperature, thus the
solvent Tm cannot be too high. On the contrary, Tg should not
be too low; our LN2 cooled cold-stage (FTIR600 stage with a
minimum temperature of ꢀ196 1C) needs to be able to keep
the quenched sample well below (B20 1C) the Tg.
Hen egg white lysozyme dissolved in water alone aggregates
upon thermal treatment above its Td; therefore an excipient is
needed. Summers and Flowers II12 showed that ionic liquids
can support reversible protein folding. Angell et al. demon-
strated the generality of this behavior in protic ionic liquids
within an optimal range of proton activity.13,14 Both groups
also demonstrated that ionic liquids can prevent protein
aggregation.5,12 Accordingly, ethylammonium nitrate (EAN)
a Department of Chemical Engineering, Princeton University,
Princeton, NJ 08544, USA. E-mail: pdebene@princeton.edu;
Fax: +1 609 258 0211; Tel: +1 609 258 5480
b Department of Chemistry and Biochemistry,
Arizona State University, Tempe, AZ 85287, USA
c International Specialty Products, 1361 Alps Road, Wayne,
NJ 07470, USA
w Electronic supplementary information (ESI) available: Further experi-
mental details, Fig. S1–S5 and Table S1. See DOI: 10.1039/b907656e
ꢁc
This journal is The Royal Society of Chemistry 2009
Chem. Commun., 2009, 4441–4443 | 4441

View Article Online
relaxation and the unfolding peak were found to be similar,
thus indicating that the quenching of lysozyme using LN2 is
able to successfully trap a fully unfolded protein. Note that
calorimetric signatures of refolding of the quenched-unfolded
sample all begin at B0 1C. Thus, changing the protein con-
centration does not cause a systematic shift in the calorimetric
onset of refolding.
Fig. 2 shows no evidence of phase separation in the
matrix. However, the matrix solution thermodynamics in the
presence of varying amounts of protein—the conditions under
which unmixing can occur and its possible role in initiating
refolding—merit detailed scrutiny and will be the subject of
future investigations.
Fourier transform IR spectroscopy is useful for protein
secondary structure detection and quantification. The best-
characterized and most-studied region is the amide I0 band16
(1600–1700 cmꢀ1 region) that arises primarily from the peptide
carbonyl stretch vibration. Since the carbonyl group forms
hydrogen bonds to peptide NH groups in a variety of different
secondary structural environments,17 the amide I0 band
involves multiple single peaks having different frequencies,
each peak being related to a specific secondary structural
element (e.g. a-helix, b-sheets). Qualitative monitoring of
structural changes can be done by following changes in
absorption at a specified frequency. Quantitative analysis
involves deconvolution of the broad amide I0 band and curve
fitting (e.g. with a superposition of Gaussian or Lorentzian
peaks). Peak areas and positions are then used to provide a
quantitative estimate of relative weights assigned to specific
secondary structural elements.18
Fig. 1 Successive DSC heating curves of 200 mg/ml lysozyme in
40 wt% sucrose and 60 wt% H2O (no EAN) and in a solution of
40 wt% EAN, 27 wt% H2O and 33 wt% sucrose (EAN) showing the
reversibility of lysozyme (unfolding–refolding) and prevention of
aggregation using EAN in this formulation with sucrose.
is added to the solvent to prevent protein aggregation as the
protein is thermally treated. This is demonstrated in Fig. 1.
Sucrose is added to the mixture to prevent ice crystallization
and so permit vitrification:15 DSC heating scans show no
evidence of an endothermic peak below 0 1C.w
As will be explained below, DSC and IR measurements
reveal an important gap in the temperature at which refolding-
related signals are manifested upon reheating. Since the
protein concentrations differ by an order of magnitude
(200 mg/ml in DSC, 20 mg/ml in IR), it was decided to test
whether lower biomolecular concentrations, with concomitant
decrease in protein–protein interactions, would shift the DSC
onset of refolding towards lower temperatures. Fig. 2 shows
DSC heating curves of quenched unfolded and quenched
folded protein for three different lysozyme concentrations.
The DSC heating curve of the quenched unfolded protein
shows an extended enthalpic relaxation with no distinct peak
Though the previous DSC-based quench-and-refold studies
used H2O,5 our IR experiments use D2O due to H2O’s over-
lapping absorbance with the amide I0 region,19 and the fact
that this absorbance exceeds that of the sample, making
accurate measurements of the signal difficult. A solvent com-
position of 2 wt% lysozyme, 26 wt% D2O, 39 wt% EAN and
33 wt% sucrose was used. In this solvent, lysozyme unfolds
at B80 1C and the solvent matrix Tg is around ꢀ100 1C
(Fig. 2). Although the solvent formulation is not exactly the
same in the DSC and IR experiments, previous calorimetric
measurements show that Tg varies by at most 10 1C in the
presence or absence of EAN.20 Quantitative analysis of IR
spectra at room temperature (25 1C) shows that the native
structure of lysozyme in our solvent mixture (42 % a-helix,
19% b-sheet, 26% turn, 13% random)w and in a neutral buffer
are very similar (40 % a-helix, 19% b-sheet, 27% turn, 14%
random).21
roughly between 0 and 60 1C, providing evidence of
non-cooperative folding. Area calculations for the enthalpic
Fig. 3 presents IR spectra that support the previously
observed5 ability to trap unfolded protein by quenching using
LN2: note, in particular, that spectra right before quench
(90 1C; thermally infolded) and right after quench (ꢀ135 1C
quench) are very similar. Once the unfolded protein is trapped
in a glassy matrix, it is slowly heated. IR scans are collected to
observe protein structural changes and intermediates as the
protein refolds to the native state. It can be seen that
the protein spectra before thermal treatment (25 1C) and after
the entire quench-and-refold process (25 1C end) are very
similar, providing spectral evidence of the reversibility of
lysozyme unfolding and subsequent refolding.
Fig. 2 Comparison of DSC upscans of solutions quenched from
30 1C (with lysozyme folded, dashed line) and quenched from 86 1C
(lysozyme unfolded, solid line): (a) 50 mg/ml lysozyme (b) 100 mg/ml
lysozyme and (c) 200 mg/ml in a solvent mixture with equal weight
each of EAN, H2O and glucose.
ꢁc
This journal is The Royal Society of Chemistry 2009
4442 | Chem. Commun., 2009, 4441–4443

View Article Online
case, then the quench-and-refold approach supplemented by
IR appears to be able to capture the formation of secondary
structural elements before energetic native contacts of the
hydrophobic core fall into place. Such energy-neutral rearrange-
ments have not been observed previously in IR. Structural
changes such as the fast hydrophobic collapse,3 can occur at
very short time scales and in most cases during other experi-
mental technique’s ‘‘dead time’’. In our studies, these very
short time scales appear to be associated with phenomena
occurring at low temperatures. T-jump kinetic studies of
hydrophobic collapse reveal a lack of activation energy
at such short time scales.23 The lack of corresponding calori-
metric signal suggests that the refolding events seen in the IR
at low temperatures are likewise energy-neutral.
In sum, we have developed a sensitive method to track the
protein intermediate structures upon folding from the fully
unfolded state back to the native state. The main advantage of
this technique is the opportunity to observe subtle structural
changes as they evolve over a continuum in time, due to the
slow kinetics found in the viscous liquid state near Tg. The
combined use of IR and DSC reveals the ability of IR to
capture early structural changes. Their possible relation to the
recently reported fast (B20 ms) hydrophobic collapse3 remains
to be established.
Fig. 3 IR spectra of lysozyme sample heated in the oven to 90 1C for
10 minutes, and then immediately quenched with LN2. The sample is
then heated from ꢀ135 1C to 25 1C at 20 1C/min with 3 minute
annealing intervals every 5 1C to collect spectra.
We thank I. Aksay for access to his IR spectrometer and the
National Science Foundation (grant no. CHE0404699 and
CHE0404714) for financial support.
Notes and references
1 C. M. Dobson, Nature, 2003, 426, 884.
2 F. Chiti and C. M. Dobson, Annu. Rev. Biochem., 2006, 75, 333.
3 L. J. Lapidus, S. Yao, K. S. McGarrity, D. E. Hertzog, E. Tubman
and O. Bakajin, Biophys. J., 2007, 93, 218.
4 W. A. Eaton, V. Munoz, S. J. Hagen, G. S. Jas, L. J. Lapidus,
E. R. Henry and J. Hofrichter, Annu. Rev. Biophys. Biomol. Struct.,
2000, 29, 327.
Fig. 4 Absorbance level at 1650 cmꢀ1 (a-helix) and 1630 cmꢀ1
(b-sheets). Error bars for 4 scans are within the size of each data point.
5 C. A. Angell and L. M. Wang, Biophys. Chem., 2003, 105, 621.
6 P. G. Debenedetti and F. H. Stillinger, Nature, 2001, 410, 259.
7 P. Privalov and N. N. Kchechinashvili, J. Mol. Biol., 1974, 86, 665.
8 P. Privalov, Adv. Protein Chem., 1979, 33, 167.
9 L. M. Wang, S. Borick and C. A. Angell, J. Non-Cryst. Solids,
2007, 353, 3829.
Fig. 4 plots the temperature dependence of absorbance
levels at chosen frequencies. It confirms the existence of
multiple conformations in the lysozyme folding process.22
Absorbance peaks at 1650 cmꢀ1 are generally associated with
a-helices, while peaks at 1630 cmꢀ1 are associated with
intramolecular b-sheets.17 At Bꢀ100 1C, a-helices start
forming. In contrast, the b-sheet content begins to decrease
at a higher temperature, Bꢀ65 1C.
10 L. J. Smith, M. Sutcliffe, C. Redfield and C. M. Dobson, J. Mol.
Biol., 1993, 229, 930.
11 S. K. Kulkarni, A. E. Ashcroft, M. Carey, D. Masselos,
C. V. Robinson and S. E. Radford, Protein Sci., 1999, 8, 35.
12 C. A. Summers and R. A. Flowers II, Protein Sci., 2000, 9, 2001.
13 N. Byrne, L. M. Wang, J. P. Belieres and C. A. Angell, Chem.
Commun., 2007, 26, 2714.
14 N. Byrne and C. A. Angell, J. Mol. Biol., 2008, 378, 707.
15 J. H. Crowe, J. F. Carpenter and L. M. Crowe, Annu. Rev.
Physiol., 1998, 60, 73.
16 J. F. Carpenter, S. J. Prestrelski and A. C. Dong, Eur. J. Pharm.
Biopharm., 1998, 45, 231.
17 W. K. Surewicz, H. H. Mantsch and D. Chapman, Biochemistry,
1993, 32, 389.
18 C. Jung, J. Mol. Recognit., 2000, 13, 325.
19 E. I. Solomon and A. B. P. Lever, Inorganic Electronic Structure
and Spectroscopy, John Wiley & Sons, Inc., New York, 1999.
20 N. Byrne and C. A. Angell, unpublished data.
21 S. Q. Luo, C. Y. F. Huang, J. F. Mcclelland and D. J. Graves,
Anal. Biochem., 1994, 216, 67.
22 A. Matagne, M. Jamin, E. W. Chung, C. V. Robinson,
S. E. Radford and C. M. Dobson, J. Mol. Biol., 2000, 297, 193.
23 M. Sadqi, L. J. Lapidus and V. Munoz, Proc. Natl. Acad. Sci.
USA, 2003, 100, 12117.
Comparison of DSC and IR data reveals an interesting
difference, in that DSC reheating curves do not show any
enthalpic relaxation (refolding) until 0 1C (see Fig. 2 and
ref. 5), while spectral shifts and qualitative changes are
observed at much lower temperatures (see Fig. 4). By contrast,
an IR upscan through the unfolding temperature range com-
plements the calorimetric findings, the a-helix signature
disappearing over the same temperature range as the heat of
unfolding is observed.w The DSC scans of Fig. 2 establish that
the difference in refolding behavior is not due to the difference
in protein concentration.
Since the integral of the exothermic deviation observed
between 0 and 60 1C is approximately equal to the unfolding
enthalpy, the energy of the protein just before the calorimetric
action starts is indeed that of the unfolded protein. If this is the
ꢁc
This journal is The Royal Society of Chemistry 2009
Chem. Commun., 2009, 4441–4443 | 4443