1-s2.0-S0001868622001567-main

Advances in Colloid and Interface Science 308 (2022) 102754
Contents lists available at ScienceDirect
Advances in Colloid and Interface Science

journal homepage: www.elsevier.com/locate/cis
Historical perspective
How do surfactants unfold and refold proteins?

Daniel E. Otzen a, b, *, Jannik Nedergaard Pedersen a, Helena Østergaard Rasmussen a, Jan
Skov Pedersen a, c, *
a
Interdisciplinary Nanoscience Center (iNANO), Aarhus University, Gustav Wieds Vej 14, 8000 Aarhus C, Denmark
b
Department of Molecular Biology and Genetics, Aarhus University, Universitetsbyen 81, 8000 Aarhus C, Denmark
c
Department of Chemistry, Aarhus University, Gustav Wieds Vej 14, 8000 Aarhus C, Denmark
A R T I C L E I N F O A B S T R A C T
Keywords: Although the anionic surfactant sodium dodecyl sulfate, SDS, has been used for more than half a century as a
SDS versatile and efficient protein denaturant for protein separation and size estimation, there is still controversy
Protein unfolding/refolding about its mode of interaction with proteins. The term “rod-like” structures for the complexes that form between
SAXS
SDS and protein, originally introduced by Tanford, is not sufficiently descriptive and does not distinguish be
ITC
Molecular dynamics
tween the two current vying models, namely protein-decorated micelles a.k.a. the core-shell model (in which
Synchrotron radiation denatured protein covers the surface of micelles) versus beads-on-a-string model (where unfolded proteins are
surrounded by surfactant micelles). Thanks to a combination of structural, kinetic and computational work
particularly within the last 5–10 years, it is now possible to rule decisively in favor of the core-shell model. This is
supported unambiguously by a combination of calorimetric and small-angle X-ray scattering (SAXS) techniques
and confirmed by increasingly sophisticated molecular dynamics simulations. Depending on the SDS:protein
ratio and the protein molecular mass, the formed structures can range from multiple partly unfolded protein
molecules surrounding a single shared micelle to a single polypeptide chain decorating multiple micelles. We also
have much new insight into how this species forms. It is preceded by the binding of small numbers of SDS
molecules which subsequently grow by accretion. Time-resolved SAXS analysis reveals an asymmetric attack by
SDS micelles followed by distribution of the increasingly unfolded protein around the micelle. The compactness
of the protein chain continues to evolve at higher SDS concentrations according to single-molecule studies,
though the protein remains completely denatured on the tertiary structural level. SDS denaturation can be
reversed by addition of nonionic surfactants that absorb SDS forming mixed micelles, leaving the protein free to
refold. Refolding can occur in parallel tracks if only a fraction of the protein is initially stripped of SDS. SDS
unfolding is nearly always reversible unless carried out at low pH, where charge neutralization can lead to su
perclusters of protein-surfactant complexes. With the general mechanism of SDS denaturation now firmly
established, it largely remains to explore how other ionic surfactants (including biosurfactants) may diverge from
this path.
1. Introduction which the hydrophilic head groups point outwards while the hydro
phobic moieties are buried in the interior [2] (Fig. 1a). Micelles are
Surfactants (surface-active agents) are small amphiphilic molecules dynamic entities in rapid equilibrium with a low concentration of
consisting of both a hydrophilic or charged part and a hydrophobic part monomeric surfactant, corresponding to the cmc. The cmc is usually
[1]. This duality gives them a particular affinity for interfacial regions from a few hundred μM to several mM, although this depends on the
between hydrophobic and hydrophilic phases, e.g. the air-water inter surfactant. Both micellar and monomeric surfactants can interact with
face. However, they can also generate their own interface in bulk water other amphiphiles such as proteins, both through polar and apolar in
by self-associating (at and above the critical micelle concentration or teractions. These interactions are both of technical, molecular and bio
cmc) to form micelles, typically shaped as ellipsoidal complexes in logical interest. For example, surfactants are widespread in Nature in the
Abbreviations: ITC, isothermal titration calorimetry; MD, molecular dynamics; SAXS, small-angle X-ray scattering; SDS, sodium dodecyl sulfate.
* Corresponding authors at: Interdisciplinary Nanoscience Center (iNANO), Aarhus University, Gustav Wieds Vej 14, 8000 Aarhus C, Denmark.
E-mail addresses: dao@inano.au.dk (D.E. Otzen), jsp@chem.au.dk (J.S. Pedersen).
https://doi.org/10.1016/j.cis.2022.102754
Received in revised form 25 July 2022;
Available online 16 August 2022
0001-8686/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
D.E. Otzen et al. Advances in Colloid and Interface Science 308 (2022) 102754
form of biosurfactants – produced by bacteria and yeasts - which interact model which is based firmly on a complementary suite of experimental
with many different biomolecules [3]. Interactions between proteins and computational approaches: the core-shell model of the SDS-
and (usually chemically synthesized) surfactants are a widespread (in denatured state, in which a surfactant micelle is surrounded or deco
some ways even defining) feature of modern society, e.g. in the deter rated by a shell of partially denatured protein. Several proteins may
gent, personal care and food industries. And from a research perspective, surround one micelle; conversely, several micelles may be surrounded
the star of the protein-surfactant show is SDS-PAGE, the most prevalent by one protein chain. It all depends on the ratio between protein and
analytical technique in protein science. Although this technique relies surfactant concentration, and the protein molecular mass. We will
critically on the ability of the anionic surfactant SDS to denature pro develop this in the coming sections.
teins as a prelude to size separation, it still remains a subject of pro
longed discussion how this denaturation actually occurs. The purpose of 2. Techniques with which to approach protein-surfactant
this review is to provide a perspective on this phenomenon, with out interactions
reaches to other types of surfactants as well. (We will largely exclude
membrane proteins from our consideration since their absolute reliance We will start with some considerations about how to study protein-
on a membrane-like or micellar environment makes it impossible to surfactant interactions. Unfortunately, the literature abounds with
study the addition of individual surfactant molecules leading up to their studies which superficially present observations about the impact of
denaturation.). Thanks to experimental and computational insights different surfactants on a protein’s structure or activity, yet fail to pro
emerging from our labs and those of others within the last 5–10 years vide any coherent insight into the nature of the mechanisms and struc
(with earlier studies leading up to this summarized in reviews in 2011 tural changes accompanying these interactions. Put simply, protein-
[4] and 2015 [5]), we believe we are now in a position to re-evaluate surfactant interactions must be studied both from the perspective of
and replace outdated views of the SDS-denatured state and settle on a the protein and the surfactant – preferably in parallel. In all cases it is
Fig. 1. Different models of micelles with and without proteins.

(a) An example of a surfactant micelle obtained by MD simulations [2]. An individual SDS molecule and a 60-molecule SDS micelle are shown. Carbon is shown as
green, hydrogen as white, oxygen as red, sulfur as orange and sodium as yellow.
(b) The pearls-on-a-string model for an unfolded protein (shown in blue, other atoms colored as in panel a). Model constructed by combining the micelle in Fig. 1a
with an extended random coil conformation of a 10 kDa protein. This model does not satisfactorily describe the complex formed between protein and SDS according
to SAXS data (see the calculated SAXS curves of similar models in Fig. 3).
(c) Protein-decorated SDS micelle. The protein has a molecular mass of 10 kDa and mainly assumes an α-helical secondary structure but has lost its tertiary fold as the
individual helices are isolated from each other. The SDS aggregation number is 70. The model is consistent with SAXS data.
(d) Schematics of the proposed changes in the architecture of the protein-decorated SDS micelle with increasing protein size. Multiple copies of small proteins can be
accommodated on one micelle, while large proteins can decorate several micelles.
(e) Flexible helix micelle model where the blue protein chain both surrounds and penetrates the central surfactant cylinder, particularly where there are hydrophobic
residues on the chain. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
2
critical to methodically titrate increasing concentrations of surfactant implementations [12]. This provides a constraint on (or control of)
into the protein solution and register and compare all transitions. The structural models provided by SAXS/SANS and molecular dynamics
different techniques fall into five classes: (MD).
(1) Protein-only: Spectroscopic techniques (Trp fluorescence and (4) The protein-surfactant complex: Small-angle scattering techniques
near- and far-UV circular dichroism (CD)) monitor protein tertiary and such as SAXS (X-ray) and SANS (neutrons) report on the size and shape
secondary structure. Note that contrary to many such claims, a of the protein-surfactant complex formed at different stages of dena
surfactant-induced increase in the far-UV CD signal in the region char turation (guided by the previous techniques), including both dynamic
acteristic of α-helices (222–208 nm) does not imply a stabilization of the and stable regions. SAXS even allows us to distinguish between protein
native state. Since the native state is the most stable and therefore all- and surfactant, without employing cumbersome and time-consuming
dominant species of the protein to start with, any spectroscopic labelling techniques, due to their different electron densities, specif
change implies a shift away from the native state and thus most likely a ically between the protein and the alkyl chain of SDS. These experi
destabilization, leading to e.g. increased occurrence of isolated α-helices. mental approaches may be complemented by MD.
Disruption of inter-helical contacts (e.g. coiled-coil structures) is typi (5) Molecular dynamics simulations of the interactions between pro
cally seen as a change in the 222/208 nm ellipticity ratio in far-UV CD tein and surfactant have emerged recently thanks to the increasing
[6] even though the overall level of helicity may be retained. There is a computational powers of simulating individual atoms. However, such
much larger CD change when β-sheet-rich or natively unfolded proteins studies are of questionable value if they fail to reproduce denaturation of
bind to SDS micelles, due to the induction of α-helical structure. α-he the protein by surfactant under ambient conditions (room temperature,
lices are often amphipathic and are therefore stabilized by the pH 7) where the protein is stable in the absence of surfactant. Failing
hydrophobic-hydrophilic interface of the micelle. this, more extreme conditions, e.g. higher temperatures, have to be
Protein stability is monitored most simply by thermal scans, which identified where the protein is stable in the absence of surfactant but
reveal if the protein has any native structure to lose, i.e. unfolds coop denatures it its presence. Docking experiments purporting to identify
eratively (as a sigmoidal curve) and with a characteristic melting tem initial binding sites for individual surfactant molecules tell us little if
perature. Ionic surfactants (IS) rarely stabilize proteins except in some anything about denaturation (and can make no claim that this is the first
cases at low concentrations where they may bind as monomers to sites stage of denaturation), simply because the protein is not denatured at
specifically found in the native state, e.g. different lipid-binding sites on this early stage. Examples of useful and less useful MD studies will be
serum albumin [7] or different regions on enzymes such as amylases [8]. presented below.
Equilibrium techniques typically only monitor protein changes up to the
formation of free micelles in solution (i.e. the apparent cmc rather than 3. Charge dependence of protein-surfactant interactions
the intrinsic cmc of the surfactant in the absence of other components
such as proteins, which is how cmc is generally defined), since there are The interactions between protein and surfactant require charge on
invariably no changes in Trp fluorescence and CD above the cmc once the surfactant but not necessarily on the protein. Let us start by dis
the protein has denatured. (This restriction is however lifted with cussing the types of surfactants and proteins that engage in these in
intramolecular distance-measurement tools such as Förster Resonance teractions. The nature of the surfactant head group is critical to the
Energy Transfer (FRET), particularly when used with single-molecule interaction. Nonionic surfactants or NIS (e.g. alkyl maltosides like
studies as we will discuss below). However, the kinetics of unfolding dodecyl maltoside (DDM) and ethylene oxides like C12E8) generally only
show a great variety of behaviors well above the cmc, indicating interact with membrane proteins (for which they form a convenient
different mechanisms and sites of attack and resulting changes in com substitute for the membrane environment [13]) and a few unstable and
plex structure as the concentration of the micelles changes (usually with dynamic globular proteins such as the apo-form of α-lactalbumin (αLA)
accompanying changes in micelle structure). [14]. This is largely because NIS’ lack of charge makes it possible for
(2) Surfactant-only: Hydrophobic fluorophores such as pyrene and them to benefit from the hydrophobicity-driven benefits of self-
ANS undergo changes in fluorescence when they transfer to a hydro association without electrostatic repulsion between head groups. In
phobic environment [9]. Thus, they report on the formation of micelles addition, the rather bulky head groups block protein access to the
(allowing them to measure cmc) as well as on the accumulation of amphiphilic interface between the micelle core and the head group re
surfactant clusters (sometimes – but often inaccurately – called hemi gion. In contrast, for IS, electrostatic repulsion between head groups is
micelles) on the protein surface which sometimes occurs. The binding of only partially mitigated by counter-ions since a large fraction of them is
surfactant by protein will increase the apparent cmc, simply because dissociated from micelles; even with counter-ions, an SDS micelle has an
binding of surfactant to proteins reduces the amount of free surfactant average charge of ~ - 0.73 per monomer [15]. Head group repulsion
molecules. However, if surfactant clusters form on the protein, they will extends the interface of the IS micelle and together with the small ionic
do so below the cmc since the protein needs to provide a more favorable head groups allows easier access to the hydrophobic part of the micelles.
binding site for surfactants than self-association, which in any case will Since self-association of NIS is generally energetically preferable to
take over at the apparent cmc. In rare cases, pyrene can also bind to protein binding, any interaction with proteins will involve micelles,
proteins with concomitant spectroscopic changes, although this does not possibly with the collaboration of monomers (present at cmc levels) to
necessarily prevent it from reporting on micelle formation alongside “soften” the protein [14].
binding [10]. Interestingly, a similar reliance on micelles to accomplish unfolding
(3) All interactions: Isothermal titration calorimetry (ITC), where is seen for carboxylate surfactants such as biosurfactants like rhamno
surfactant at high concentration is titrated into a solution of protein, lipid (RL) and sophorolipid (SL), whose lack of denaturation potency is
follows the enthalpy changes associated with demicellisation, as well as attributed to their delocalized charge and complex hydrophobic chains
surfactant binding to proteins and protein unfolding. This registers the which probably favors pure micelles [16]. Like other NIS, RL unfolds
otherwise silent binding of small numbers of (sometimes individual) SDS αLA below the cmc but unfolding kinetics accelerate around the
molecules and typically reveals a much richer and more complex series apparent cmc as the micelles can start to contribute [17], while SL only
of changes than the simple studies in (1). Importantly, when performed unfolds αLA above the apparent cmc [18].
at several protein concentrations, they provide the stoichiometry of In contrast to the rather self-reliant NIS, proteins represent an
binding of surfactant molecules to the protein (as well as the concen alternative environment for SDS and other IS in which to build up
tration of non-bound surfactants) at any given stage in the interactions clusters of surfactant molecules at sub-cmc concentrations (though often
[11]. Complex enthalpograms can be analyzed objectively to provide smaller than micelles in solution). Due to the ability to associate the
robust thermodynamic parameters with recent software anionic sulfate head groups with the proteins, SDS can benefit from self-
3
assembling their alkyl moieties while reducing the penalty of electro large, one protein can even cover several micelles. This is illustrated in
static repulsion. Viewed in this light, the role of the protein is to stabilize Fig. 1d. It is important to ensure that SDS is present at sufficiently high
individual SDS micelles by decorating their surface, engaging with ionic concentration to achieve this state. A more restricted study of βLG-SDS
head groups to reduce repulsion. Protein denaturation becomes a interactions using SAXS and DLS only examined βLG-SDS complexes up
consequence as the protein is extended to maximize the interaction area. to an early stage in the SDS titration and failed to reach conditions which
As will be discussed further below, the alternative model of micelle- lead to core-shell structures [31]. Typically the log of the rate of
decorated polypeptide chains (pearls-on-a string model, Fig. 1b), in unfolding scales linearly with [SDS] in this concentration range [4,33],
which an unfolded protein chain is surrounded by micelles rather than indicating that the gradual accretion of SDS micelles can be viewed as a
the reverse, not only lacks direct experimental support but also fails to single process where the different levels of decoration fundamentally
provide any additional stability to the micelle or rationalize unfolding of represent the same type of denaturation.
the extensive parts of the protein not in contact with micelles. Micelle
stabilization is also seen with the so-called liprotides: here cis-unsatu 5. Experimental evidence for the protein-decorated micelle
rated fatty acids like oleic acid (OA), which normally form complex model
bilayer structures at neutral pH and only form micelles at pH above 11
[19], are able to form micelles down to pH as low as 6 [20] when sta Support for the protein-decorated micelle (core-shell, Fig. 1c) from
bilized by a shell of partially unfolded proteins such as αLA [21]. scattering experiments has been available for a long time, although
It is important to note that protein-head group interactions may be quantitative modelling of the scattering data has emerged more recently
promoted by, but are not absolutely reliant on, complementary elec [31,32]. In 1990 Ibel et al. [34] used SANS contrast variation to study
trostatic interactions (i.e. SDS binding to Lys/Arg and avoiding Asp/ PRibAnt-isomerase/IndGroP-synthase saturated with SDS. By Indirect
Glu). Protein electrostatics do play a role and individually bound SDS Fourier Transformation (IFT) of the scattering data (which provides an
molecules may well contact Lys and Arg residues in the native protein orientationally averaged self-correlation function of the complexes) at
[22]. Indeed, the protein S6 is more rapidly denatured by SDS below pH different contrasts, the authors derived a decorated micelle model. This
5 where Asp and Glu are presumably protonated in a micellar envi was supported by clever use of per-deuterated protein (scattering
ronment [23]. Further, acetylation of the 18 Lys residues in carbonic differently from SDS) and H2O-D2O mixtures to render either protein or
anhydrase (which still leaves 9 unmodified Arg) lead to greater SDS SDS invisible and thus localize one component relative to the other in
resistance [24] and molecular dynamics simulations show that Lys res the complex. However, at that time a theoretical expression for the
idues are early points of contact for SDS during unfolding of ACBP [25]. scattering of such structures was not available, precluding quantitative
On the other hand, acetylation of the 7 Lys in ubiquitin (leaving 4 Arg) modelling of the SANS data. In a pioneering work from 1995, Samso
has little effect on SDS-induced unfolding (although slightly fewer SDS et al. [35] combined cryogenic transmission electron microscopy (cryo-
molecules bind initially) [26], and – perhaps most importantly - a pro TEM) and SAXS to study the complex between SDS and several proteins.
tein completely lacking ionizable residues (i.e. only with charges at the TEM micrographs clearly revealed micelle-like structures and proteins
two termini) is unfolded by SDS in a completely conventional manner: It larger than 20 kDa were able to connect several micelles. Qualitative
unfolds below the cmc, has a high SDS binding capacity, shows multiple analysis of the SAXS data using IFT were also consistent with the
(though mainly endothermic due to lack of electrostatic interactions) consensus core-shell model, possibly with some hydrophobic residues
transitions by ITC, unfolds rapidly and forms a core-shell complex with penetrating into the core of the SDS micelles. Nevertheless, this study
SDS [27]. The rapid unfolding kinetics indicate that electrostatic in did not yet have access to a geometric model to fit directly to the data.
teractions are not even critical in reducing activation barriers to Quantitative modelling was introduced in our study of the
unfolding. Rather the data highlight the importance of having access to complexation of SDS with the protein ACBP [32], in which we used
an amphiphilic interface through small SDS head groups. Monte Carlo generated points to present the SDS micelle core and shell
as well as protein. We used this approach as it allows generating rather
4. Protein denaturation involves the formation of a core-shell complex structures for which analytical expressions for the scattering
structure intensity are not available. Additional restraints on concentration of
components, scattering length densities of the various components (SDS
As summarized in previous reviews [4,5], protein denaturation by tail and head, and protein) and partial molecular volumes were
SDS can be regarded as a multi-step process. Although details vary from employed in the model, ensuring that the models were physically
protein to protein, broadly speaking the first stage involves accumula reasonable. Strikingly, all the derived models that provided good fits to
tion of a few individual SDS molecules on the protein (which, depending the SAXS data were protein-decorated micelles, and some of the models
on conditions such as pH, can lead to dimerization of the native state at low SDS:protein ratio had an extended shell suggesting accumulation
with surfactant acting as glue [28–30]) while subsequent stages lead to of several proteins around a single micelle. Since the structure of the
protein-stabilized micelles. These micelles are often smaller than the complexes were determined to be ellipsoidal, it is possible to derive
free micelles; SDS micelles have an aggregation number of around 70 on analytical expressions for the scattering intensity. This was done in a
their own but in complex with e.g. αLA as little as 43 SDS molecules are later study on myoglobin and α-lactalbumin [31], where restraints on
required to form a stable micelle [31]. Similar numbers are seen with concentrations and scattering length densities as well as partial molec
ACBP (36–38) [32] and ubiquitin (40) [28]. For the core-shell structure ular volumes were also employed.
(Fig. 1c) that describes the structure of the complex, the shell thickness Although SANS notably employs many useful labelling strategies to
(including protein, SDS head groups and bound water) increases to highlight certain components, label-free experiments in SAXS can be
about 10 Å upon protein binding (compared to 5 Å for the pure micelle), very informative by exploiting natural contrast within protein-SDS
indicating that proteins associate closely to the micelle in an unfolded complexes. X-ray contrasts arise from electron density differences.
state and partly replace water in the shell. The hydrophobic core of the Since the hydrophobic (carbon-rich) core of alkyl chains has a lower
complexes has a very similar size as in pure micelles; this is evidence for electron density than all conventional buffers, its excess scattering
insignificant penetration of protein into the micellar core [31]. At low length density is negative. The sulfate head groups and the counter ions
SDS:protein ratios, several protein (ACBP [32], αLA [31] or ubiquitin are electron rich, and consequently give rise to positive excess scattering
[28]) molecules can combine around a single micelle, with only parts of length density. When examining pure SDS micelles, this difference be
the molecule in contact with the micelle. With increasing [SDS], the tween alkyl core and head group-counter ion shell and the globular and
number of proteins per micelle decreases until eventually each micelle highly symmetric shape give rise to a deep minimum in the SAXS in
will only be surrounded by one protein; if the protein is sufficiently tensity at a scattering vector modulus, q, around 0.09 Å− 1 (broken black
4
line in Fig. 2) followed by a broad secondary maximum in the region conditions, protein and SDS have similar contrast and therefore cannot
0.09–0.35 Å− 1 (see e.g. [31,36,37] and Fig. 2, black arrow). The SAXS be unequivocally distinguished. As a consequence, modelling could not
data for micellar SDS can be described on absolute intensity scale by an rule on whether the protein resides on (decorates) the surface or goes
ellipsoidal core-shell model with an oblate (flattened disc) core and a through the core of the micelles. The SANS data were modelled as
thin shell with a width of about 5 Å [31], where concentrations, scat globular micelles forming a fractal aggregate with a relatively high
tering length densities, and partial molecular volumes of the compo number of micelles per aggregate. For SAXS, on the other hand, a
nents are used as restraints. The protein is also quite electron rich and polypeptide chain going through the SDS micelle (Fig. 3) gives a scat
therefore also has a positive excess scattering length density as tering curve very different from that of decorated micelle models. In the
compared to water and conventional buffers. The SAXS intensity profile calculation of the model intensities, two models were considered for the
of SDS-protein complexes at SDS saturation resembles that of pure SDS pearls-on-a-string structure: One where the peptide chain is linear and
micelles with a deep minimum (Fig. 2, red broken line) and a broad another one where it adopts random coil configurations outside the SDS
maximum for slightly higher q (see e.g. [31,36,37] and Fig. 2, red micelle. The calculation is based on the formalism described in [42,43]
arrow). The main difference is at low q, where the intensity is much and uses a contour length of 4.0 Å per residue [44], a Kuhn length of
larger for SDS-protein complexes due to the large positive scattering peptide chains of ~15 Å [45] and an average diameter of an unfolded
length contribution of protein. The retention of a deep minimum and polypeptide of 3.25 Å. The results of these model calculations are pre
broad maximum in protein-SDS complexes shows that the symmetry and sented in Fig. 3 for two pearls-on-a-string models and the core-shell
structure of the formed complexes are very similar to that of pure SDS decorated micelles. Both pearls-on-a-string models fail to reproduce
micelles, suggesting a protein-decorated micelle. This has been the deep minimum in the scattering intensity and the pronounced
confirmed using detailed modelling with a model on absolute intensity maximum at higher q, which are observed experimentally. The absence
scale consisting of a core-shell structure, where the core consists of the of this minimum arises from the protein’s positive scattering length
alkyl chains and the shell of sulfate head groups, counter ions, protein density within the alkyl core, which has a negative scattering length
and hydrating water. Using again molecular and concentration re density. This redistribution of scattering material give rise to a large
straints, this model gives rise to excellent fits to the SAXS data (see e.g. change in the micellar maximum at high q in the scattering curve for
[28,31,36–39]). Importantly, the shortest core dimensions of the SDS- both models. Therefore, the pearls-on-a-string models can be ruled out.
protein complexes are very similar to that of pure SDS micelles (they The other alternative model, the flexible helix micelle model, consists
are independent of the SDS aggregation number, since this radius is of a flexible cylindrical SDS micelle with the protein wrapped around it
given by the length of the alkyl chain), showing that there is very little as a more or less regular helix (Fig. 1e) [46]. The regular helix is
penetration of protein into the core. incompatible with SAXS and SANS data since it would give rise to a
Bragg-like peak in the measured q range. The cylindrical shape can
6. Alternative models for protein-surfactant complexes easily be ruled out for complexes with low molecular mass proteins, as
the ellipsoidal model provides perfect fits. For higher molecular mass
Other models proposed in the literature fail to reproduce charac proteins, it is more difficult to rule out a cylindrical model with protein
teristic scattering features of protein-surfactant complexes. Two alter on the surface by simple arguments, since the scattering from several
native complex models have been suggested, both of which have connected decorated micelles is similar to that of a cylinder. However,
features in common with the protein-decorated micelle model. In the connected protein-decorated ellipsoidal micelles are the only way to
pearls-on-a-string model (Fig. 1b), micelles assemble on an unfolded explain the experimental evidence from cryo-TEM by Samso and col
polymer chain, so that the polypeptide chain passes through the center leagues [35], the contrast-variation SANS study by Ibel and colleagues
of the micelles (see e.g. [40,41]). However, these studies employed SANS [34] and the oscillatory behavior of the pair-distance distribution
on non-deuterated protein and SDS in D2O buffers. Under these
10-1
10 -1
I(q) [cm-1]
I(q) [cm-1]
10-2
10-2
10-3
0.01 0.1 1 10-3

0.01 0.1 1
q [Å-1]
q [Å-1]
Fig. 2. SAXS data for pure SDS micelles (black points) and for saturated
lysozyme-SDS complexes in the absence of free SDS micelles (red points). The Fig. 3. Calculation of theoretical SAXS curves for a decorated core-shell com
protein concentration is 5 mg/mL and the SDS concentration is 25 mM for both plex (black), a pearls-on-a-string model, where the protein is linear (red), and a
samples. Broken lines indicate the position of the deep minimum (indicative of similar model, where the protein forms two random coils outside the micelle
shape symmetry) and the arrows indicate the broad micelle-associated bump. (blue). The micelles contain 40 SDS molecules and the protein has a molecular
Data from [80]. (For interpretation of the references to color in this figure mass of 10 kDa. (For interpretation of the references to color in this figure
legend, the reader is referred to the web version of this article.) legend, the reader is referred to the web version of this article.)
5
function from IFT due to a high occurrence of distances between the 7. Time-resolved SAXS and molecular dynamics reveal an
individual decorated micelles for complexes of SDS and high molecular asymmetric attack strategy
mass proteins [30]. Despite this strong evidence in favor of the protein-
decorated ellipsoidal micelle model, closer inspection suggests that Just as the structure of the protein-SDS complex changes with
some cases in the literature, originally modelled as pearls-on-a-string, in increasing [SDS], it also develops over time. By combining SAXS with
fact are better explained as protein-decorated cylindrical models stopped-flow kinetics at a synchrotron source, which allows accumula
[40,41,47]. Indeed, there are even models claiming to follow the pearls- tion of spectra down to the millisecond time-scale, we have recently
on-a-string model which in reality are modelled as core-shell structure, uncovered different steps in the formation of the core-shell complex
since the protein strands going through the micelles are not included in between SDS and β-lactoglobulin (βLG) [36] as well as αLA [37] (sum
the model [48]. marized in Fig. 4). Due to the high protein concentrations required, SDS
By itself, SDS undergoes such a globule-to-cylinder transition in salt is used at super-cmc concentrations (though at protein:SDS stoichiom
[49], so it is very reasonable to expect the structure of the decorated etries consistent with distinct transitions in SDS binding seen by ITC)
micelles in the complexes to be cylindrical. In [33] as well as in the and therefore engages the protein as preformed micelles. The first spe
works of Chen and Teixiera [47] and Kohlbrecher and colleagues [40], cies detected after a few ms are multiple-associated/aggregated protein-
the size of the protein-SDS and protein‑lithium dodecyl sulfate com SDS complexes which subsequently dissociate to single micelle-protein
plexes indicates the presence of several proteins per complex, as judged complexes. Initially the protein is asymmetrically distributed around
from the ratio of the low-q intensity and the intensity at higher q, before the micelle – which is perhaps unsurprising given that the micelle will
the strong decay due to the cross-section/local structure (this ratio gives have to attack the protein from one side anyway. Over the next few
an approximate estimate of how many times the length is larger than the seconds, the protein gradually repositions more evenly around the
cross-section diameter). A multi-protein complex would be unlikely for micelle as it unfolds. This basic mechanism is observed at the highest
decorated ellipsoidal micelles, since a gain in mixing entropy would be measured protein:SDS ratios (1:22–1:63). At the lowest measured pro
obtained for complexes that only involve one protein. tein:SDS ratio (1:12) the protein was only partly unfolded and the
It should be mentioned that there have been objections to the gradual repositioning of the protein around SDS was not visible.
decorated micelle models for BSA [50]. The argument was that due to its Gratifyingly, important features of this denaturation process are
many internal disulfide bonds, the protein does not have sufficient recapitulated when we perform MD simulations of the unfolding of our
flexibility to decorate a micelle. However, the experimental evidence for model protein ACBP by SDS [25]. In practice we need to present SDS to
the protein-decorated micelle model from scattering methods is over the protein in a non-micellar state, otherwise no denaturation is seen
whelming, so despite the many internal links, BSA clearly has enough within the time span of the all-atom simulations (1 μs). (Similarly, a
flexibility to wrap around micelles in a partly unfolded state. Other computational study with interleukin 8 with 60-molecule SDS micelles
disulfide-bonded proteins such as αLA [31], lysozyme [51] and the at 300 K failed to see unfolding in silico within 1 μs [55]). Due to MD
cellulase Cel7b [38] also form core-shell complexes. computational constraints, the 80:1 SDS:ACBP molar ratio translates to
Unfortunately, other models than the established protein-decorated 520 mM SDS and 6.5 mM protein, which is of course far from experi
micelle model are still being put forward, for example for the model mental conditions. However, the high concentration likely also accel
protein ubiquitin. Proposed models include an unfolded protein chain erates the unfolding process and allows it to complete within a 1 μs run.
decorated with individual SDS molecules [52,53] (surmised from indi Micellization forms part of the denaturation process which is initiated by
rect methods such as capillary electrophoresis and NMR) or even native- the binding of clusters of 10–20 SDS molecules to different parts of the
like structures surrounded by SDS micelles [54]. The latter model is protein, followed by binding and uptake of more surfactant molecules
based on MD simulations which in any case are hampered by the fact until these micelles merge to a mature micelle with the protein disor
that the protein is not denatured by SDS in silico at 27 ◦ C; within the time dered on the surface. Initial binding is driven by a combination of
span of MD, unfolding only takes place when the protein is pre- electrostatics (Lys residues) and hydrophobic contacts; the sulfate
incubated at 97 ◦ C after which it is exposed to SDS which under these groups also disrupt internal hydrogen bonds to replace them with
situations appear to adopt non-micellar structures. Gratifyingly, we have hydrogen bonds to the sulfate group. Over time, the protein rearranges
recently used SAXS to vindicate experimentally the core-shell model for on the micelle surface to reduce the asymmetry of the complex.
SDS-ubiquitin complexes [28]. There are further twists to the behavior Importantly, predicted SAXS spectra of structures obtained by these
of ubiquitin in SDS. Analysis of different complexes formed by step-wise simulations match well with experimental SAXS spectra, and the initial
addition of SDS demonstrates that binding of a small number of SDS sites of attack are consistent with previous hydrogen-exchange [56] and
molecules (3–4 per protein) promotes protein dimerization, while mutational [33] studies, making this a robustly validated study. In
higher [SDS] transforms the system to a decorated 40-SDS micelle [28]. Fig. 5, SAXS data for around 56 SDS molecules per ACBP are compared
This is accompanied by loss of secondary and tertiary structure as the to the simulation results for different time points (200, 400, 600, 800,
protein unfolds more and more to cover more and more micellar surface 1000 ns). For the simulations, the complexes coexist with small micelle/
area, leading to a smaller number of ubiquitin molecules per micelle. aggregates of SDS so that the complexes contain 63–66 SDS molecules.
Furthermore, the kinetics of unfolding of ubiquitin accelerate abruptly To compensate for the difference in aggregation numbers, the contrast of
around 60 mM SDS and undergo a switch in the amplitude of the signal the SDS heads groups/counterions, and protein, respectively, were
(Otzen et al., unpublished results). optimized relative to that of the micelle core. A structure factor was also
Remarkably, the same core-shell structure is seen for apo-αLA and included to model the complex-complex interaction effects in the data
myoglobin when SDS is replaced by the biosurfactant RL [31]. RL forms apparent as a broad maximum at small scattering vectors. The fits to the
smaller micelles (nagg = 11, much less than the typical 70 for SDS) and SAXS data were compelling, with small adjustments of the contrasts,
this small number is maintained when it forms complexes with αLA. except for 800 ns, where the complex is less symmetric. (Note that the fit
Overall only 0.48 g RL bind per αLA, which together with its larger shown in fig. 12a of [25] is an incorrect curve from an analytical core-
molecular weight makes RL bind only αLA at ca. 25% of the molar ratio shell model fit. The correct curve is now provided in Fig. 5.) Interest
of SDS, but this does not prevent it from forming the same core-shell ingly, when these simulations are repeated using sodium hexadecyl
structures. sulfate (SHS), a surfactant with four more carbon atoms in the alkyl
chains, the same pattern is seen though on a much shorter time-scale:
SHS forms larger attacking clusters and there is more rapid loss of
native contacts and expansion of the protein around the growing SHS
micelle. Similar asymmetric attack and redistribution is observed in a
6
Fig. 4. Steps involved in the unfolding of β-lactoglobulin in SDS (top) and refolding from the SDS-denatured state into an excess of the nonionic surfactant C12E8
(bottom), based on synchrotron-SAXS experiments. Figure adapted from [36]. Note that for simplicity the micelles are depicted as spherical but they are in fact oblate
ellipsoidal.
water. Formation of decorated micelles is also seen in a simulation of

SDS unfolding of the β-sheet protein I27 [58], where individual SDS
0.1
molecules rather than pre-formed micelles are used to ensure unfolding

(consistent with the related protein TII27’s ability to unfold in SDS
below the cmc [59]). In silico I27 is first unfolded at 100 ◦ C before
incubating with SDS at 27 ◦ C. Reminiscent of ACBP, first individual SDS
0.01 molecules bind via cationic residues and clear the way for micelles to
form and unfold the protein. The protein is not symmetrically positioned
I(q) [cm ]
-1
around the micelles after 6 μs of simulation, but this may reflect a lack of
SAXS data
200 ns
equilibration, as I27 unfolds remarkably slowly in SDS with half-lives
400 ns between 2 h (2 mM SDS) and 2 min (500 mM SDS) [59].
600 ns Thus, despite the limitations of current MD studies, contours are
800 ns
0.001
1000 ns emerging of a general model for the mechanisms of attack by SDS from
MD simulations in combination with stopped-flow kinetics studies
[14,33,39] and underpinned by experimental SAXS spectra of end-stage
complexes. Attack by individual SDS molecules (always present, also
above the cmc) paves the way for micelle formation and subsequent
0.01 0.1 1
q [Å-1] unfolding and extension of the protein around the emerging micelles.
Once increased computational powers allow us to extend to more
Fig. 5. SAXS data for ACBP for complexes with 56 SDS molecules and one experimentally tractable time scales and avoid the need for high con
ACBP. The curves are calculated for the MD results at time points 200–1000 ns
centrations of monomeric SDS, heating or coarse-grained simulations, it
as indicated in the figure. With small adjustment of the scales of the contrasts
may be possible to establish even more robust unfolding pathways. The
and inclusion of a structure factor to the described concentration effects, the fits
to the data are quite good, except for 800 ns, where the complexes are less
next step will be to incorporate actual rates of unfolding as constraints
symmetric. Figure based on data presented in [25]. into the simulations, inspired by recent developments in modelling
folding of proteins from denatured state ensembles [60].
coarse-grained MD simulation of the formation of core-shell liprotides
8. Super-cmc studies: stopped-flow kinetics and single-molecule
by αLA together with preformed micelles of OA (reflecting the extremely
techniques
low cmc of OA) [57], which likewise has been validated by experimental
SAXS data. The only twist compared to SDS is that αLA penetrates more
Our SAXS analysis is best at probing structures formed at [SDS]
deeply into the core of the OA micelles, either because of the lower
where the protein complexes gradually become saturated with SDS; at
charge of the carboxylate group of OA compared to SDS or simply
higher [SDS], protein-free micelles will gradually dominate the SAXS
because we here use coarse grain representations and force fields that
signal and make it increasingly difficult to assign contributions from the
generally overestimate the attraction between molecules that are not
protein. Stopped-flow kinetics at super-cmc concentrations provide
7
insight into changes in the mechanism of unfolding induced by changes molecule, which forms a 9–10 Å thick shell around the core, and it is
in micellar concentrations (since the monomer concentration remains reasonable to assume that the SDS-unsaturated system contains com
essentially constant), e.g. reducing the rate of unfolding of ACBP at plexes with several S6 molecules per complex, as observed for other
higher concentrations, presumably due to blocking of unfolding by proteins of similar mass. Therefore the original schematic figure in [61],
additional micelles [33] (but accelerating it in the case of ubiquitin). For although compatible with the FRET data, has been redrawn as protein-
proteins such as S6, unfolding can be accelerated, presumably due to decorated micellar structures with termini close to each other in order to
shifts in micelle structure towards more cylindrical species, although take subsequent SAXS data into account.
this effect is not seen nearly as clearly with other proteins like CI2 [23].
However, this approach does not provide direct information about the 9. pH-dependence of SDS unfolding
protein structural changes accompanying this binding.
To escape this impasse, we turned to single-molecule Förster Reso Unfolding by SDS is strongly pH-dependent and can be accompanied
nance Energy Transfer (smFRET) analyses using a variant of S6 labelled by aggregation through super clustering. Although proteins lacking
with donor and acceptor fluorophores at either terminus [61]. This ionizable side chains can still be unfolded by SDS, some proteins with a
approach gives us information both on the compactness of individual S6 full collection of ionizable side chains resist SDS denaturation, largely
molecules and their population-wide distribution, the distribution of co- due to a high activation barrier to unfolding, i.e. high kinetic stability.
existing species and the dynamics of their exchange, although it does not This class of stable proteins is dominated by rigid β-sheet proteins [62]
provide any information about the associated SDS. Remarkably, we saw like outer membrane proteins [63], superoxide dismutase and trans
multiple transitions between different co-existing species not only below thyretin [64], but also includes enzymes like the lipase from Thermo
and around saturation but also at several hundred mM SDS, far above myces lanuginosus (TlL). TlL is designed to be activated by interfacial
saturation (summarized in Fig. 6). Between 0.4 and 1.5 mM SDS, as the regions; accordingly low concentrations of both NIS and IS strongly
protein underwent unfolding according to Trp fluorescence and far-UV increase activity, but impede it at higher concentrations, presumably by
CD, S6 formed an expanded native-like state coinciding with a more blocking the active site [65]. The high activation barrier also impedes
expanded FRET state, both with very rapid (μs-ms) internal dynamics refolding of the urea- and pH-denatured state in the presence of SDS,
but which interconverted between each other on the second time-scale, leading to hysteresis [66]. Remarkably, this hysteresis is highly pH
consistent with bulk stopped-flow kinetics. Between 1.5 and 10 mM SDS, dependent and a combination of ITC, SAXS has very recently revealed an
the two FRET labels approach each other, implying that S6 either forms enormous range of different species of complexes occurring at different
a more compact state or the two labels are brought together within the pH values [30] (Fig. 7). Hysteresis is most pronounced around pH 8.0
confines of the core-shell structure. Above 50 mM SDS, this state coexists where TlL is both intrinsically stable and has an overall negative charge
with a more expanded state (i.e. where the FRET-molecules are further (− 12), with isolated patches of positive potential. Once folded (in the
apart) with ms time-scale interconversion rates. Given the extremely low absence of SDS), TlL remains folded at pH 8.0 even at high [SDS]. This is
(pM) protein concentrations, we are unlikely to be observing FRET be confirmed by its ability to undergo cooperative unfolding in thermal
tween different protein molecules. The data suggest that the core-shell scans although the melting temperature decreases from 74 ◦ C to around
model could in fact be an ensemble of different protein molecules, all 44 ◦ C [65].
of which are attached to micelles but with different levels of compact The decline in stability of this folded state is also seen by hydrogen/
ness. To confirm that the SDS-S6 complexes conform to the general deuterium exchange detected by mass spectrometry (HDX-MS) which
scenario that we have outlined in the present review with formation of reveals an overall higher level of deuterium exchange in SDS, although
protein-decorated micelles at SDS saturation, we have performed SAXS the overall protection pattern of the native state is maintained [30]. ITC
measurements on a sample with 4.0 mg/mL S6 and 20 mM SDS, where at pH 8.0 reveals several binding steps, of which the first one has a
the complexes are saturated with SDS with little free SDS in solution. counterintuitive negative binding number (− 10), while subsequent
The SAXS data could be modelled by the conventional core-shell model steps lead to the accumulation of 10–25 SDS molecules, well below the
with a prolate ellipsoidal core with 37 SDS molecules and a single S6 maximal capacity of ca. 143 SDS (i.e. 1.4 g SDS/g TlL). Modelling of the
Fig. 6. Complexity of SDS-mediated unfolding of S6 assessed by single-molecule FRET studies. Shown are the four stages of S6 unfolding in SDS: The native state
(upper left, shown as cartoon), the unfolded and expanded state (lower, left, and middle), the compact state (middle, right), and the unfolded state depicted as
decorated micelles in a core-shell model (right) [61].
8
Fig. 7. Overview of TlL:SDS complexes formed at pH 4.0, 6.0 and 8.0 at increasing [SDS]. The hemimicelle and cmc regions are shown in pink and light blue,
respectively, and are determined by pyrene fluorescence. The purple, orange, and blue line show development of fluorescence at 342 nm, activity, and the median of
deuteration (from HDX-MS data), respectively, as a function of [SDS]. The concentrations denote [SDS] for each structure. Note that different [SDS] are measured at
each pH value based on the characteristic transition points determined by ITC, which vary with pH. Data from [30,66]. (For interpretation of the references to color
in this figure legend, the reader is referred to the web version of this article.)
SAXS spectra at pH 8.0 reveals that at relatively low [SDS] , where net charge remains negative (− 5), there are no longer any contiguous
apparent negative binding numbers are found, there are in fact about 18 patches of negative charge. The protein still forms native-state dimers at
SDS molecules bound at one side of the dimer at the interface between low and intermediate [SDS] (Fig. 7, pH 6.0, 1.4 and 3.2 mM), but twice
the two TlL molecules (Fig. 7, pH 8.0, 4.6 mM). Increasing levels of as many SDS (39) are bound to each lipase and this is enough to even
dimerization with increasing protein concentration will reduce the tually tip the balance towards denaturation. At higher [SDS], lipase
number of accessible SDS binding sites and can thus lead to apparent unfolds around the micelle to form classical core-shell structures,
negative binding numbers. In addition, calculation of binding stoichi initially one micelle per chain and subsequently two per chain, leading
ometries is complicated by overlap of endothermic and exothermic to 109 SDS per lipase (Fig. 7, pH 6.0, 7.3 and 10.8 mM).
contributions to the enthalpy whose relative weight and [SDS] depen Interestingly, both RL and sophorolipid (SL) support dimerization of
dence change with protein concentration. At higher [SDS], TlL remains native TlL at pH 6.0 but not at pH 8.0 [29], and rather than destabilizing
dimeric but initially stabilizes one small micelle at intermediate [SDS] TlL, they significantly increase its melting temperature. Thus, the
which then increases in size until saturation is reached (according to ITC decreased electrostatic repulsion at pH 6.0 presumably helps this native-
data). Thus, the native state of TlL stabilizes the micelle, while the level dimerization, no doubt stabilized by hydrophobic surfactant in
micelle conversely renders TlL more dynamic and thus less stable, but teractions. Similar interactions may be at play when low [SDS] promote
not unstable enough to unfold at room temperature. Besides showing an native-state dimerization of the charge-less protein EXG before unfold
overall destabilization of the natively folded structure, HDX-MS com ing it into core-shell complexes at higher [SDS] [39].
plements SAXS by providing residue-specific information about protein- Returning to TlL and SDS, one might assume that a further decrease
SDS interactions. For instance, HDX-MS revealed that certain parts of in pH would lead to direct core-shell formation but the behavior at pH
TlL are actually protected by SDS, i.e. they exchange less upon addition of 4.0 is more complicated. Apart from the last step in the titration of SDS
SDS. These parts include the lid region and two other helices which all into lipase, most of the transitions measured by ITC lead to visibly turbid
contain features such as hydrophobic patches and positive charges. solutions, yet remarkably SAXS is able to distinguish between these
These binding sites constitute anchor points for SDS [30] and are also different aggregates. The aggregates form under conditions where
consistent with SDS’ ability to interact with the lid region according to 20–40 SDS bind per lipase, effectively neutralizing the net charge of
activity measurements [65]. +18. The aggregates can be described as different combinations of
A decrease in pH to 6.0 radically changes this situation. Although the compact clusters joined by random-flight segments into superclusters
9
(Fig. 7, pH 4.0, 0.6, 1.9 and 3.3 mM SDS); eventually these thin out to study the dependence of this process on MFSDS and the underlying
become linearly connected random-flight structures of multiple (on mechanisms involved. ITC is not particularly useful for this since the
average 17) protein-decorated micelles joined by short strings of pro concentration of monomeric SDS is extremely low in mixed micelles; for
teins (Fig. 7, pH 4.0, 4.6 mM SDS) and finally a dual-micelle complex example, DDM has a cmc of ca. 0.17 mM (almost 50-fold lower than SDS
(Fig. 7, pH 4.0, 8.7 mM SDS) similar to the one observed at pH 6.0. An in water) and most SDS is therefore taken up into micelles already at
earlier study of the unfolding of the cellulase Cel7b in SDS at pH 4.2 also these low concentrations. This more or less eliminates SDS’ denaturing
revealed formation of superclusters at low [SDS] which eventually give activities as monomer (thus ruling out ITC studies) and reduces the
way to a core-shell complex with 20–40 SDS/protein and multiple mi situation to an assessment of the properties of the mixed micelle.
celles per protein [38]. Fortunately, SAXS is highly appropriate for these studies since we
Similar transitions involving aggregates in different sized clusters, can deconvolve the scattering signal for (relatively) complex mixtures
which eventually give way to isolated core-shell complexes, are seen for into individual components, allowing us to distinguish between dena
egg white lysozyme [51]. The enzyme is highly basic (pI = 10.7) and has tured proteins in micelles, empty micelles and native proteins. The
a charge of +8 at pH 7, meaning that charge neutralization can occur at process can be monitored by complementary spectroscopic techniques
neutral pH and does not require acidic conditions unlike Cel7b and to gauge the extent of refolding. With this approach, we have studied the
lipase. The only difference is that aggregation and dissolution occurs refolding of SDS-unfolded proteins such as βLG, lysozyme and serum
over a relatively narrow SDS concentration range as the protein quickly albumin by addition of NIS (C12E8 or DDM) [80]. There is a simple
becomes sufficiently anionic to resolubilize. Conversely, the highly transition from denatured proteins in core-shell complexes to “naked”
acidic protein pepsin (pI 3.24) is so negatively charged at neutral pH that native protein and mixed micelles, which coincides with the regain of
it is highly resistant to SDS unfolding [67], while easily precipitated by a both secondary and tertiary structure. This directly demonstrates that
cationic surfactant like CTAB at pH values above 3 [68]. The structures SDS can be removed from protein and that SDS does not (necessarily)
of these complexes remain unresolved. It has been known for a long time trap proteins irreversibly in a non-native state – at least not at neutral
(reviewed in [69]) that low concentrations of SDS can induce protein pH. When the same refolding approach is repeated with Cel7b at pH 4.2,
aggregation through either neutralization of charge at low pH or by refolding competes with aggregation due to charge neutralization,
stabilizing aggregation-prone partially folded states at neutral pH, leading to a mixture of complexes of different sizes, only a small fraction
which over time can form amyloid-like intermolecular contacts [70]. of which is the micelle-free native protein [38]. We avoid aggregation if
The behavior of both lipase and cellulase (both of which resist SDS we bypass the unfolded state by addition of mixed micelles with the
denaturation at neutral pH) adds another twist to this: although it is same MFSDS as the end-point level of the refolded state. Thus, it may be
possible to alter SDS affinity through changes in pH, this can easily lead concluded that the siphoning off of SDS from the Cel7b-SDS complex
to aggregation as well as denaturation and the formation of a very takes place in a gradual fashion so Cel7b transitions through a region
intricate network of bonded structures, in which the key element re corresponding to SDS:protein ratios that lead to aggregation [38]. Of
mains stabilization and decoration of micelles of different sizes. course this approach presupposes that the protein does not interact with
NIS; if that is the case (as with αLA), then we merely see a transition from
10. From unfolding to refolding: the use of mixed micelles an SDS-denatured state to a NIS-bound state, which for αLA can be
modelled as a dimer held together by 13–20 DDM molecules [80].
Mixed micelles may be used to uncover parallel tracks of fast and Although the refolding seems like a prosaic affair when monitored at
slow stripping and folding involved in unfolding to refolding processes. equilibrium, a stopped-flow synchrotron SAXS study reveals a more
Unfolding of proteins by SDS is not an irreversible process. SDS can be complex mechanism (Fig. 4). When SDS-denatured βLG is turbulently
removed efficiently in a “strip-away” process by adding SDS- mixed with micelles of C12E8, there is an almost instantaneous diverging
sequestering agents such as cyclodextrins [71] which readily take up response for the denatured population: some proteins are immediately
different surfactants [72,73]; this is so rapid that it can be used to stripped of SDS (leaving the naked denatured state) and rapidly refold to
measure intrinsic (rapid) rates of refolding [74]. However, it is difficult the native state, while a comparably sized population remains bound to
to control the extent of SDS removal this way, particularly at low [SDS] SDS and is only relieved of bound surfactant molecules more slowly,
(which invariably represents the interesting region). An alternative is to leading to a slower refolding process. This leads to a biphasic accumu
let SDS “taste its own medicine” by diluting it out with NIS which readily lation of native-state structure seen by both near- and far-UV CD [36].
incorporate SDS into mixed micelles [75]. This approach has several The lower the MFSDS, the greater the proportion of SDS-free denatured
advantages. Scientifically, it allows us to tune the potency of the mixed species that initiate fast-track folding. Thus, the rate-limiting step in
micelles by varying the mole fraction of SDS, MFSDS (= [SDS]/([SDS] + practice becomes the removal of SDS from the protein complex to allow
[NIS])), and investigate its effect on protein structure. Varying MFSDS rapid folding.
from 0 to 1 has been adopted as an approximately linear denaturation
scale for membrane proteins [76–78] though one should be aware of 11. Membrane proteins in SDS: glimpses of a denatured state
potential limitations at very low MFSDS [79] due to incomplete mixing of
surfactants with different cmc values [75]. Globular proteins also unfold Integral membrane proteins present a very different situation from
more slowly in mixed micelles than in pure SDS micelles [59]. globular proteins when it comes to studying their interactions with SDS.
An additional benefit of mixed micelles is their application rele Since they are embedded in a hydrophobic environment in their native
vance. Detergent mixtures typically contain both anionic surfactants and state, they cannot exist as stable folded monomers in aqueous buffer in
NIS, since NIS are better at solubilizing oils and organic dirt and are less the absence of lipid vesicles or functionally equivalent substitutes such
sensitive to divalent metal cations. Based on all these considerations, it as NIS micelles. Therefore, it is not possible to delineate the stepwise
was a triumphant vindication to demonstrate that addition of NIS like addition of SDS to a “naked” membrane protein to monitor unfolding.
C12E8 and DDM to SDS-denatured proteins can restore the native state of Consequently we shall limit ourselves to a few comments about the
the protein, or simply prevent unfolding if present with SDS from the impact of SDS on these proteins. It is possible to titrate SDS into mem
beginning [80]. Similarly it allows the maintenance of enzyme activity brane proteins folded in NIS micelles of e.g. DDM [76]. This typically
in the presence of the anionic surfactant LAS [81]. Block copolymers leads to a single equilibrium step [77,84] (rarely two [76]), though ki
[82,83] can likewise sequester SDS, allowing them to refold HSA from netic studies may pick up more steps [85]. The SDS-denatured state of an
SDS [83] or take up small molecules from HSA in complex with CTAB integral membrane protein becomes the most appropriate version of the
without compromising protein structural integrity (based on CD mea denatured state and therefore of particular interest for membrane pro
surements) [82]. Given these undisputed facts, it is more interesting to tein folding studies, although ingenious studies that trap membrane
10
proteins in unfolded states using steric trapping also can provide insight 13. Future perspectives: is SDS a typical surfactant?
into lipid-solubilized denatured state ensembles [86]. Yet we still have
little direct insight into the structural properties of this state. Is it too Although we have significantly advanced our understanding of the
naïve to expect the secondary structure of the native state (closely structures and mechanisms of formation of protein-surfactant com
packed α-helices) to be reconfigured as isolated helices, i.e. simply an plexes, these are invariably biased by the dominant role played by SDS.
expanded and more dynamic version of the original native state? Studies Our glimpses of the complex between αLA and RL suggest that the same
on bacteriorhodopsin using spin-labels and EPR indicate a relatively broad outline is followed by other IS, in which the protein stabilizes
compact state with fluid helical arrangements and no organized struc micelles from an early stage, leading to the generic core-shell structure.
ture [87]. Clearly this state is fundamentally different from the core- Nevertheless, it will be interesting to study how other surfactants behave
shell structure adopted by globular proteins, reflecting the need for for comparison in order to put this further into perspective. For example,
embedding directly in the membrane. This will likely be an exciting area will unfolding of proteins by mixtures of anionic surfactants and NIS be
to explore further with combinations of smFRET, SAXS and SANS. dominated by regular-sized micelles or will they allow formation of
smaller micellar species along the way as is suggested by MD simulations
12. Application of other techniques to study protein-surfactant with SDS? Or is the question irrelevant because the cmc of mixed mi
interactions celles is much lower than that of pure anionic surfactants so bulk mi
celles in practice dominate the reaction? And do cationic surfactants
We find it pertinent to consider the use of new applications which lead to the same core-shell structures as anionic surfactants? The simple
could widen the field of protein-surfactant investigations. Although the answer to this latter question may be yes: BSA unfolding by cationic
SAXS technique has proven itself to be a nearly perfect tool for structure trimethyl ammonium bromide-type surfactants of different chain
determination for surfactant-protein complexes due to the natural large lengths has been proposed to occur via decorated micelles [96]
difference in contrast between micellar core and protein, SANS provides (although the authors refer to the structure as pearls-on-a-string struc
additional unique possibilities including contrast variation and selective ture, they are in practice decorated micelles). Nevertheless, that study
isotope labelling. Since the first work by Ibel and coworkers [34], there identifies complexes with multiple proteins and micelles, significantly
have been few examples of the application of contrast variation SANS for larger than those seen with SDS. This may reflect a balance between
structure determination of protein-detergent complexes [88,89]. Per charge neutrality (which together with longer alkyl chains encourages
forming the measurements with one of the components per-deuterated, larger aggregates; BSA has a net charge of − 18 at pH 7) and mixing
i.e. protein or SDS, and for several different ratios of H2O and D2O gives entropy (favoring smaller complexes).
unique information about the relative position of protein and SDS in the This leads on to the general question: How do other surfactants differ
complexes. Simultaneous fitting of such SANS data taken at different from SDS in their denaturation mechanisms? For example, amino acid
contrast with a molecular and concentration constrained core-shell surfactants such as dodecyl sarcosinate show low toxicity and high de
model [90,91] would therefore offer a unique possibility to investigate gradability, but have so far only been subjected to very limited analysis
possible partial penetration of the protein into the core region of the SDS [97]. Similarly, gemini surfactants, which weld together two individual
micelle. Due to the large efforts with deuteration and restricted SANS surfactant molecules [98] through adjustable spacer lengths [99] have
beam time, future studies will probably have to be limited to well- also attracted significant interest but the field is in need of more sys
selected representative examples of SDS-undersaturated complexes tematic studies to address their mode of action properly, rather than
which contain several low-molecular mass proteins, the corresponding simply limiting studies to e.g. activity measurements [100,101]. Activity
saturated complexes with one protein per complex, and high-molecular measurements using insoluble substrate like starch should take into
mass proteins which connect several protein-decorated micelles. account how surfactants can affect substrate. Thus the claimed activa
The enormous advances made by cryo-EM over the past decade may tion of amylase by SDS and CTAB [102] is most likely an indirect effect
also provide further insight into the structures of individual protein- on substrate and requires soluble substrate for investigation; when we
surfactant complexes. The technique has already been applied to study amylase activity using soluble hexaose we see no increase in ac
investigate complexes in which SDS acts as a conventional molecular tivity though SDS clearly stabilizes the protein against denaturation [8].
ligand rather than a micelle-forming denaturant, e.g. activation of he Other indirect effects include adsorption of NIS onto lignin in cellulose
mocyanin [92] (though without direct identification of individual SDS degradation; this also has a positive effect since it prevents unspecific
molecules). Protein conformations in bona fide core-shell complexes are binding and inactivation of cellulases [103,104].
dynamic by virtue of their denaturation which makes it meaningless to In summary, this review set out to present the latest developments in
expect atomic-level resolution, but single-particle analysis and classifi the field and to establish that the core-shell model is the only plausible
cation schemes may well sketch out the full range of protein-micelle model for SDS/protein interactions, in which one or more partially
configurations, particularly when combined with the average struc unfolded polypeptide chains decorate the surfaces of one or more sur
tures provided by SAXS and the diversity of distance measurements factant micelles. This model is supported both by experimental and
obtained from smFRET. computational studies, and a possible mechanism of formation over time
We are less confident about prospects for mass spectrometry. Native has been delineated. Importantly, complex formation is reversible
state MS has been able to analyze complexes of micelles and membrane around neutral-to-basic pH and folded protein can be recovered by the
proteins for the last decade [93] but has mainly been used to investigate addition of NIS; at lower pH values charge neutralization can trap the
bound lipid molecules such as lipid “plugs” [94]. SDS is traditionally protein-surfactant species in insoluble but highly organized superclu
incompatible with MS analysis due to its strong signal suppression [95] sters. With this as a framework, a primary aim of the field is now to
and it remains unclear whether this can be overcome in native state MS generalise the core-shell model and extend the range of surfactants
to obtain a quantitative analysis of the full range of SDS:protein stoi (including many in common use) and the range of target proteins.
chiometries and complex sizes. However, HDX-MS can provide impor
tant indirect information about specific SDS binding sites and/or CRediT authorship contribution statement
protected sites due to protein-protein interactions or even allosteric
changes, as long as SDS is removed without significant back-exchange Daniel E. Otzen: Conceptualization, Formal analysis, Funding
prior to MS analysis [30]. acquisition, Investigation, Methodology, Project administration, Super
vision, Writing – original draft, Writing – review & editing. Jannik
Nedergaard Pedersen: Formal analysis, Investigation, Writing – review
& editing. Helena Østergaard Rasmussen: Formal analysis,
11
Investigation, Writing – review & editing. Jan Skov Pedersen: Formal [20] Pedersen JN, Frislev HS, Pedersen JS, Otzen DE. Using protein-fatty acid
complexes to improve vitamin D stability. J Dairy Sci 2016;99:7755–67.
analysis, Funding acquisition, Investigation, Methodology, Project
[21] Kaspersen JD, Pedersen JN, Hansted JG, Nielsen SB, Sakthivel S, Wilhelm K, et al.
administration, Supervision, Writing – review & editing. Generic structures of liprotides, complexes between partially denatured 27
proteins and oleic acid: a fatty acid core with a shell of disordered proteins.
ChemBioChem 2014;18:2693–702.
Declaration of Competing Interest [22] Yonath A, Podjarny A, Honig B, Sielecki A, Traub W. Crystallographic studies of
protein denaturation and renaturation. 2. Sodium dodecyl sulfate induced
structural changes in triclinic lysozyme. Biochemistry 1977;16:1418–24.
The authors declare that they have no known competing financial [23] Otzen DE. Protein unfolding in detergents: effect of micelle structure, ionic
interests or personal relationships that could have appeared to influence strength, pH, and temperature. Biophys J 2002;83:2219–30.
[24] Gitlin I, Gudiksen KL, Whitesides GM. Peracetylated bovine carbonic anhydrase
the work reported in this paper. (BCA-Ac18) is kinetically more stable than native BCA to sodium dodecyl sulfate.
J Phys Chem B 2006;110:2372–7.
Acknowledgements [25] Poghosyan AH, Schafer NP, Lyngso J, Shahinyan AA, Pedersen JS, Otzen DE.
Molecular dynamics study of ACBP denaturation in alkyl sulfates demonstrates
possible pathways of unfolding through fused surfactant clusters. Protein Eng Des
We are grateful for funding over the years to research on protein- Sel 2019;32:175–90.
surfactant interactions. D.E.O. thanks the Novo Nordisk Foundation [26] Shaw BF, Schneider GF, Arthanari H, Narovlyansky M, Moustakas D, Durazo A,
et al. Complexes of native ubiquitin and dodecyl sulfate illustrate the nature of
for instrument support (grants NNF14OC0011283 and hydrophobic and electrostatic interactions in the binding of proteins and
NNF14OC0012953) along with Augustinus Fonden (grant 13-0970) and surfactants. J Am Chem Soc 2011;133:17681–95.
project support from the Independent Danish Research Fund (grants [27] Hojgaard C, Sorensen HV, Pedersen JS, Winther JR, Otzen DE. Can a charged
surfactant unfold an uncharged protein? Biophys J 2018;115:2081–6.
4005-00479 and 12-126186), the Danish Ministry of Science, Technol [28] Mortensen HG, Otzen DE, Pedersen JS. Ubiquitin forms conventional decorated
ogy and Innovation (BIOPRO) and the Danish Innovation Foundation micelle structures with sodium dodecyl sulfate at saturation. J Colloid Interface
(4105-00008B-DFORT). J.S.P. thanks the Independent Danish Research Sci; 2021. 596, 233-244.
[29] Madsen JK, Kaspersen JD, Andersen CB, Pedersen JN, Andersen KK, Pedersen JS,
Fund (grants 4002-00479B, 8021-00133B and 9041-00075B) and the
et al. Glycolipid biosurfactants activate, dimerize, and stabilize thermomyces
Lundbeck foundation (grant R164-2013-15574). J.S.P. and D.E.O. also lanuginosus lipase in a pH-dependent fashion. Biochemistry-Us 2017;56:
acknowledge funding from the Novo Nordisk Foundation (grant 4256–68.
NNF21OC0071946). [30] Rasmussen HØ, Wollenberg DTW, Wang H, Andersen KK, Oliveira CL,
Jørgensen CI, et al. The changing face of SDS denaturation: complexes of
Thermomyces lanuginosus lipase with SDS at pH 4, 6 and 8. J Colloid Interface Sci
References 2022;614:214–32.
[31] Mortensen HG, Madsen JK, Andersen KK, Vosegaard T, Deen GR, Otzen DE, et al.
Myoglobin and alpha-Lactalbumin form smaller complexes with the biosurfactant
[1] Nakama Y. Chapter 15 - Surfactants. In: Sakamoto K, Lochhead RY, Maibach HI,
Rhamnolipid Than with SDS. Biophys J 2017;113:2621–33.
Yamashita Y, editors. Cosmetic Science and Technology. Amsterdam: Elsevier;
[32] Andersen KK, Oliveira CLP, Larsen KL, Poulsen FM, Callisen TH, Westh P, et al.
2017. p. 231–44.
The role of decorated SDS micelles in sub-cmc protein denaturation and
[2] MacKerell AD. Molecular dynamics simulation analysis of a sodium dodecyl
association. J Mol Biol 2009;391:207–26.
sulfate micelle in aqueous solution: decreased fluidity of the micelle hydrocarbon
[33] Andersen KK, Otzen DE. How chain length and charge affect surfactant
interior. J Phys Chem 1995;99:1846–55.
denaturation of ACBP. J Phys Chem B 2009;113:13942–52.
[3] Otzen DE. Biosurfactants and surfactants interacting with membranes and
[34] Ibel K, May RP, Kirschner K, Szadkowski H, Mascher E, Lundahl P. Protein-
proteins: same but different? Biochim Biophys Acta 2017;1859:639–49.
decorated micelle structure of sodium-dodecyl-sulfate–protein complexes as
[4] Otzen DE. Protein-surfactant interactions: a tale of many states. Biochim Biophys
determined by neutron scattering. Eur J Biochem 1990;190:311–8.
Acta 2011;1814:562–91.
[35] Samso M, Daban JR, Hansen S, Jones GR. Evidence for sodium dodecyl sulfate/
[5] Otzen DE. Proteins in a brave new surfactant world. Curr Op Coll Interface
protein complexes adopting a necklace structure. Eur J Biochem 1995;232:
Science 2015;20:161–9.
818–24.
[6] Cooper TM, Woody RW. The effect of conformation on the CD of interacting
[36] Pedersen JN, Lyngso J, Zinn T, Otzen DE, Pedersen JS. A complete picture of
helices: a theoretical study of tropomyosin. Biopolymers 1990;30:657–76.
protein unfolding and refolding in surfactants. Chem Sci 2020;11:699–712.
[7] Deep S, Ahluwalia JC. Interaction of bovine serum albumin with anionic
[37] Jensen GV, Pedersen JN, Otzen DE, Pedersen JS. Multi-step unfolding and
surfactants. Phys Chem Chem Phys 2001;3:4583–91.
rearrangement of alpha-lactalbumin by SDS revealed by stopped-flow SAXS.
[8] Madsen JK, Pihl R, Møller ALB, Tranberg Madsen A, Otzen DE, Andersen KK. The
Front Mol Biosci 2020;7:125.
anionic biosurfactant rhamnolipid does not denature industrial enzymes. Front
[38] Rasmussen HO, Enghild JJ, Otzen DE, Pedersen JS. Unfolding and partial
Microbiol 2015;6:292.
refolding of a cellulase from the SDS-denatured state: from beta-sheet to alpha-
[9] Ananthapadmanabhan KP, Goddard ED, Turro NJ, Kuo PL. Fluorescence probes
helix and back. Biochim Biophys Acta Gen Subj 2019;1864:129434.
for critical micelle concentration. Langmuir 1985;1:352–5.
[39] Højgaard C, Sørensen HV, Pedersen JS, Winther JR, Otzen DE. Can a charged
[10] Hansted JG, Wejse PL, Bertelsen H, Otzen DE. Effect of protein-surfactant
surfactant unfold an uncharged protein? Biophys J (as Biophysical Letter) 2018;
interactions on aggregation of β-lactoglobulin. Biochim Biophys Acta 2011;1814:
115:2081–6. 28.
713–23.
[40] Mehan S, Aswal VK, Kohlbrecher J. Tuning of protein-surfactant interaction to
[11] Nielsen AD, Arleth L, Westh P. Analysis of protein-surfactant interactions - a
modify the resultant structure. Phys Rev E 2015;92.
titration calorimetric and fluorescence spectroscopic investigation of interactions
[41] Guo XH, Zhao NM, Chen SH, Teixeira J. Small-angle neutron scattering study of
between Humicola insolens cutinase and an anionic surfactant. Biochim Biophys
the structure of protein/detergent complexes. Biopolymers 1990;29:335–46.
Acta 2005;1752:124–32.
[42] Svaneborg C, Pedersen JS. A formalism for scattering of complex composite
[12] Tidemand FG, Zunino A, Johansen NT, Hansen AF, Westh P, Mosegaard K, et al.
structures. I. Applications to branched structures of asymmetric sub-units. J Chem
Semi-empirical analysis of complex ITC data from protein-surfactant interactions.
Phys 2012;136:104105.
Anal Chem 2021;93:12698–706.
[43] Svaneborg C, Pedersen JS. A formalism for scattering of complex composite
[13] Neumann J, Klein N, Otzen DE, Schneider D. Folding energetics and
structures. II. Distributed reference points. J Chem Phys 2012;136:154907.
oligomerization of polytopic α-helical transmembrane proteins. Arch Biochem
[44] Ainavarapu SR, Brujic J, Huang HH, Wiita AP, Lu H, Li L, et al. Contour length
Biophys 2014;564:281–96.
and refolding rate of a small protein controlled by engineered disulfide bonds.
[14] Otzen DE, Sehgal P, Westh P. α-lactalbumin is unfolded by all classes of
Biophys J 2007;92:225–33.
detergents but with different mechanisms. J Colloid Interface Sci 2009;329:
[45] Correction for Kohn et al., Random-coil behavior and the dimensions of
273–83.
chemically unfolded proteins, PNAS 2004 101:12491-12496. Proc Natl Acad Sci
[15] Bales BL. A definition of the degree of ionization of a micelle based on its
U S A 2005;102:14475.
aggregation number. J Phys Chem B 2001;105:6798–804.
[46] Lundahl P, Greijer E, Sandberg M, Cardell S, Eriksson K-O. A model for ionic and
[16] Madsen JK, Pihl R, Moller AH, Madsen AT, Otzen DE, Andersen KK. The anionic
hydrophobic interactions and hydrogen bonding in sodium dodecyl
biosurfactant rhamnolipid does not denature industrial enzymes. Front Microbiol
sulfate–protein complexes. Biochim Biophys Acta 1986;873:20–6.
2015;6:292.
[47] Chen SH, Teixeira J. Structure and fractal dimension of protein-detergent
[17] Andersen KK, Otzen DE. Denaturation of α-lactalbumin and myoglobin by the
complexes. Phys Rev Lett 1986;57:2583–6.
anionic biosurfactant Rhamnolipid. Biochim Biophys Acta 2014;1844:2338–45.
[48] Gelamo EL, Itri R, Alonso A, da Silva JV, Tabak M. Small-angle X-ray scattering
[18] Andersen KK, Vad BS, Roelants S, van Bogaert IA, Otzen DE. Weak and saturable
and electron paramagnetic resonance study of the interaction of bovine serum
protein-surfactant interactions in the denaturation of apo-alpha-lactalbumin by
albumin with ionic surfactants. J Colloid Interface Sci 2004;277:471–82.
acidic and lactonic sophorolipid. Front Microbiol 2016;7.
[49] Jensen GV, Lund R, Gummel J, Narayanan T, Pedersen JS. Monitoring the
[19] Pedersen JN, Frislev HKS, Pedersen JS, Otzen DE. Structures and mechanisms of
transition from spherical to polymer-like surfactant micelles using small-angle X-
formation of liprotides. Biochim Biophys Acta Proteins Proteomics 1868;2020:
ray scattering. Angew Chem Int Ed 2014;53:11524–8.
140505.
12
[50] Takeda K, Moriyama Y. Comment on the misunderstanding of the BSA− SDS [77] Paslawski W, Kristensen JV, Lillelund O, Schafer N, Baker R, Urban S, et al.
complex model: concern about publications of an impractical model. J Phys Chem Cooperative folding of a polytopic α-helical membrane protein involves a
B 2007;111:1244. compact N-terminal nucleus and non-native loops. Proc Natl Acad Sci U S A 2015;
[51] Sun Y, Filho PLO, Bozelli JC, Carvalho J, Schreier S, Oliveira CLP. Unfolding and 112. 7978-7983. 30.
folding pathway of lysozyme induced by sodium dodecyl sulfate. Soft Matter [78] Hong H. Toward understanding driving forces in membrane protein folding. Arch
2015;11:7769–77. Biochem Biophys 2014;564:297–313.
[52] Schneider GF, Shaw BF, Lee A, Carrilho E, Whitesides GM. Pathway for unfolding [79] Hong H, Blois TM, Cao Z, Bowie JU. Method to measure strong protein-protein
of ubiquitin in sodium dodecyl sulfate, studied by capillary electrophoresis. J Am interactions in lipid bilayers using a steric trap. Proc Natl Acad Sci U S A 2010;
Chem Soc 2008;130:17384–93. 107:19802–7.
[53] Shaw BF, Schneider GF, Arthanari H, Narovlyansky M, Moustakas D, Durazo A, [80] Kaspersen JD, Sondergaard A, Madsen DJ, Otzen DE, Pedersen JS. Refolding of
et al. Complexes of native ubiquitin and dodecyl sulfate illustrate the nature of SDS-unfolded proteins by nonionic surfactants. Biophys J 2017;112:1609–20.
hydrophobic and electrostatic interactions in the binding of proteins and [81] Russell GL, Britton LN. Use of certain alcohol ethoxylates to maintain protease
surfactants. J Am Chem Soc 2011;133:17681–95. stability in the presence of anionic surfactants. J Surfactant Deterg 2002;5:5–10.
[54] Jafari M, Mehrnejad F, Rahimi F, Asghari SM. The molecular basis of the sodium [82] Mora AK, Basu A, Kalel R, Nath S. Polymer-assisted drug sequestration from
dodecyl sulfate effect on human ubiquitin structure: a molecular dynamics plasma protein by a surfactant with curtailed denaturing capacity. Phys Chem
simulation study. Sci Rep 2018;8:2150. Chem Phys 2019;21:7127–36.
[55] Dominguez H. Interaction of the interleukin 8 protein with a sodium dodecyl [83] Mondal R, Ghosh N, Paul BK, Mukherjee S. Triblock-copolymer-assisted mixed-
sulfate micelle: a computer simulation study. J Mol Model 2017;23. micelle formation results in the refolding of unfolded protein. Langmuir 2018;34:
[56] Kragelund BB, Knudsen J, Poulsen FM. Local perturbations by ligand binding of 896–903.
hydrogen deuterium exchange kinetics in a four-helix bundle protein, acyl [84] Faham S, Yang D, Bare E, Yohannan S, Whitelegge JP, Bowie JU. Side-chain
coenzyme A binding protein (ACBP). J Mol Biol 1995;250:695–706. contributions to membrane protein structure and stability. J Mol Biol 2004;335:
[57] Nedergaard Pedersen J, Frederix PLTM, Skov Pedersen J, Marrink SJ, Otzen DE. 297–305.
Role of charge and hydrophobicity in liprotide formation: a molecular dynamics [85] Otzen DE. Folding of DsbB in mixed micelles: a kinetic analysis of the stability of a
study with experimental constraints. Chembiochem 2018;19. 263-271. 29. bacterial membrane protein. J Mol Biol 2003;330:641–9.
[58] Winogradoff D, John S, Aksimentiev A. Protein unfolding by SDS: the microscopic [86] Gaffney KA, Guo R, Bridges MD, Muhammednazaar S, Chen D, Kim M, et al. Lipid
mechanisms and the properties of the SDS-protein assembly. Nanoscale 2020;12: bilayer induces contraction of the denatured state ensemble of a helical-bundle
5422–34. membrane protein. Proc Natl Acad Sci U S A 2022;119.
[59] Nielsen MM, Andersen KK, Westh P, Otzen DE. Unfolding of β-sheet proteins in [87] Krishnamani V, Hegde BG, Langen R, Lanyi JK. Secondary and tertiary structure
SDS. Biophys J 2007;92:3674–85. of bacteriorhodopsin in the SDS denatured state. Biochemistry 2012;51:1051–60.
[60] Brotzakis ZF, Vendruscolo M, Bolhuis PG. A method of incorporating rate [88] Lin J-M, Lin T-L, Jeng U-S, Huang Z-H, Huang Y-S. Aggregation structure of
constants as kinetic constraints in molecular dynamics simulations. Proc Natl Alzheimer amyloid b(1-40) peptide with sodium dodecyl sulfate as revealed by
Acad Sci 2021;118:e2012423118. small-angle X-ray and neutron scattering. Soft Matter 2009;5:3913–9.
[61] Krainer G, Hartmann A, Bogatyr V, Nielsen J, Schlierf M, Otzen DE. SDS-induced [89] Zhang XL, Penfold J, Thomas RK, Tucker IM, Petkov JT, Bent J, et al. Self-
multi-stage unfolding of a small globular protein through different denatured assembly of hydrophobin and hydrophobin/surfactant mixtures in aqueous
states revealed by single-molecule fluorescence. Chem Sci 2020;11:9141–53. solution. Langmuir 2011;27:10514–22.
[62] Manning M, Colón W. Structural basis of protein kinetic stability: resistance to [90] Cusack S, Ruigrok R, Krygsman P, Mellema JE. Structure and composition of
sodium dodecyl sulfate suggests a central role for rigidity and a bias toward beta- influenza virus. A small-angle neutron scattering study. J Mol Biol 1985;186(3):
sheet structure. Biochemistry 2004;43:11248–54. 565–82.
[63] Dornmair K, Kiefer H, Jähnig F. Refolding of an integral membrane protein. [91] Pedersen JS. Structure of clathrin-coated vesicles from small-angle scattering
OmpA of Escherichia coli. J Biol Chem 1990;265:18907–11. experiments. Eur Biophys J 1993;22:79–95.
[64] Xia K, Zhang S, Bathrick B, Liu S, Garcia Y, Colon W. Quantifying the kinetic [92] Cong Y, Zhang Q, Woolford D, Schweikardt T, Khant H, Dougherty M, et al.
stability of hyperstable proteins via time-dependent SDS trapping. Biochemistry Structural mechanism of SDS-induced enzyme activity of scorpion hemocyanin
2012;51:100–7. revealed by electron cryomicroscopy. Structure 2009;17:749–58.
[65] Mogensen JE, Sehgal P, Otzen DE. Activation, inhibition and destabilization of [93] Barrera NP, Di Bartolo N, Booth PJ, Robinson CV. Micelles protect membrane
Thermomyces lanuginosus lipase by detergents. Biochemistry 2005;44:1719–30. complexes from solution to vacuum. Science 2008;321:243–6.
[66] Wang H, Andersen KK, Sehgal P, Hagedorn J, Westh P, Borch K, et al. pH [94] Liko I, Degiacomi MT, Mohammed S, Yoshikawa S, Schmidt C, Robinson CV.
regulation of the kinetic stability of the lipase from Thermomyces lanuginosus. Dimer interface of bovine cytochrome c oxidase is influenced by local
Biochemistry 2013;52:264–76. posttranslational modifications and lipid binding. Proc Natl Acad Sci U S A 2016;
[67] Nelson CA. The binding of detergents to proteins. I. The maximum amount of 113:8230–5.
dodecyl sulfate bound to proteins and the resistance to binding of several [95] Rundlett KL, Armstrong DW. Mechanism of signal suppression by anionic
proteins. J Biol Chem 1971;246:3895–901. surfactants in capillary electrophoresis-electrospray ionization mass
[68] Chakraborty T, Chakraborty I, Moulik SP, Ghosh S. Physicochemical studies on spectrometry. Anal Chem 1996;68:3493–7.
pepsin− CTAB interaction: energetics and structural changes. J Phys Chem B [96] Saha D, Ray D, Kohlbrecher J, Aswal VK. Unfolding and refolding of protein by a
2007;111:2736–46. combination of ionic and nonionic surfactants. Acs Omega 2018;3:8260–70.
[69] Otzen DE. Amyloid formation in surfactants and alcohols: membrane mimetics or [97] Rudra S, Dasmandal S, Mahapatra A. Binding interaction of sodium-N-dodecanoyl
structural switchers? Curr Prot Peptide Sci 2010;11:355–71. sarcosinate with hemoglobin and myoglobin: physicochemical and spectroscopic
[70] Giehm L, Oliveira CLP, Christiansen G, Pedersen JS, Otzen DE. SDS-induced studies with molecular docking analysis. J Colloid Interface Sci 2017;496. 267-
fibrillation of α-synuclein: an alternative fibrillation pathway. J Mol Biol 2010; 277. 31.
401:115–33. [98] Sharma T, Dohare N, Kumari M, Singh UK, Khan AB, Borse MS, et al. Comparative
[71] Rozema D, Gellman SH. Artificial chaperone-assisted refolding of carbonic effect of cationic gemini surfactant and its monomeric counterpart on the
anhydrase B. J Biol Chem 1996;271:3478–87. conformational stability and activity of lysozyme. RSC Adv 2017;7:16763–76.
[72] Sehgal P, Sharma D, Larsen KL, Wimmer R, Otzen DE, Doe H. Influence of beta- [99] Adak S, Datta S, Bhattacharya S, Banerjee R. Role of spacer length in interaction
cyclodextrin on the mixed micellization process of sodium dodecyl sulfate and between novel gemini imidazolium surfactants and Rhizopus oryzae lipase. Int J
sodium lauroyl sarcosine and formation of inclusion complexes. J Dispers Sci Biol Macromol 2015;81:560–7.
Technol 2008;29:128–33. [100] Zhang J, Zhang J. Study on the interaction of alkaline protease with main
[73] Sehgal P, Sharma M, Larsen KL, Wimmer R, Doe H, Otzen DE. Interactions of surfactants in detergent. Colloid Polym Sci 2016;294:247–55.
γ-cyclodextrin with the mixed micelles of anionic surfactants and their inclusion [101] Magalhães SS, Alves L, Sebastião M, Medronho B, Almeida ZL, Faria TQ, et al.
complexes formation. J Dispers Sci Technol 2008;29:885–90. Effect of ethyleneoxide groups of anionic surfactants on lipase activity. Biotechnol
[74] Otzen DE, Oliveberg M. A simple way to measure protein refolding rates in water. Prog 2016;32:1276–82.
J Mol Biol 2001;313:479–83. [102] Antony N, Balachandran S, Mohanan PV. Effect of surfactants on catalytic activity
[75] Sehgal P, Mogensen JE, Otzen DE. Using micellar mole fractions to assess of diastase α-amylase. J Surfactant Deterg 2014;17:703–8.
membrane protein stability in mixed micelles. Biochim Biophys Acta 2005;1716: [103] Kristensen JB, Börjesson J, Bruun MH, Tjerneld F, Jørgensen H. Use of surface
59–68. active additives in enzymatic hydrolysis of wheat straw lignocellulose. Enzyme
[76] Lau FW, Bowie JU. A method for assessing the stability of a membrane protein. Microb Technol 2007;40:888–95.
Biochemistry 1997;36:5884–92. [104] Eriksson T, Börjesson J, Tjerneld F. Mechanism of surfactant effect in enzymatic
hydrolysis of lignocellulose. Enzyme Microb Technol 2002;31:353–64.
13

1-s2.0-S0001868622001567-main

Uploaded by

Copyright:

Available Formats

1-s2.0-S0001868622001567-main

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1-s2.0-S0001868622001567-main

Uploaded by

Copyright:

Available Formats

Advances in Colloid and Interface Science 308 (2022) 102754

Contents lists available at ScienceDirect

Advances in Colloid and Interface Science

How do surfactants unfold and refold proteins?

Fig. 1. Different models of micelles with and without proteins.

0.01 0.1 1 10-3

water. Formation of decorated micelles is also seen in a simulation of

molecules rather than pre-formed micelles are used to ensure unfolding

You might also like