Subscriber access provided by - Access paid by the | UCSB Libraries
Article
Structure and dynamic of heteroprotein coacervates Paulo De Sa Peixoto, Guilherme M Tavares, Thomas Croguennec, Aurélie Nicolas, Pascaline Hamon, Claire Roiland, and Saïd Bouhallab Langmuir, Just Accepted Manuscript • DOI: 10.1021/acs.langmuir.6b01015 • Publication Date (Web): 29 Jun 2016 Downloaded from http://pubs.acs.org on June 30, 2016
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Langmuir is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
1 2
Structure and dynamic of heteroprotein coacervates
3
4
5
6
Paulo D.S. Peixoto,† Guilherme M. Tavares,†,‡ Thomas Croguennec,†
7
Aurélie Nicolas,† Pascaline Hamon,† Claire Roiland,§ Saïd Bouhallab †,*
8 9 10 11
†
STLO, UMR1253, INRA, Agrocampus Ouest, 35000, Rennes, France
12
‡
Laboratory of Research in Milk Products, Universidade Federal de Viçosa, BR-36570 Viçosa,
13
Brazil
14
§
15
74205, France
Institut des Sciences Chimiques de Rennes, UMR CNRS 6226, Université de Rennes 1 - CS
16 17 18 19 20 21 22
*corresponding author: Tel: +33 223485742; email:
[email protected] 23
ACS Paragon Plus Environment
1
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 26
24
Abstract
25
Under specific conditions, mixing two oppositely charged proteins induces a liquid-liquid phase
26
separation. The denser phase, or coacervate phase, can be potentially applied as a system to protect
27
or encapsulate different bioactive molecules with a broad range of food and/or medical applications.
28
Optimization of the design and efficiency of such systems require a precise understanding of the
29
structure and the equilibrium of the nano-complexes formed within the coacervate. Here, we report
30
on the nano-complexes and the dynamics of the coacervates formed by two well-known, oppositely
31
charged proteins β-lactoglobulin (pI ≈ 5.2) and lactoferrin (pI ≈ 8.5). Fluorescence recovery after
32
photo bleaching (FRAP) and solid state NMR experiments indicate the co-existence of several
33
nano-complexes as primary units for the coacervation. This is the first (to the best of our
34
knowledge) evidence on the occurrence of an equilibrium between quite instable nano-complexes in
35
the coacervate phase. Combined with in silico docking experiments, these data support the fact that
36
coacervation in the present heteroprotein system depends not only on the structural composition of
37
the coacervates but also on the association rates of the proteins forming the nano-complexes.
38 39 40
Keywords: Heteroprotein coacervation, β-lactoglobulin, lactoferrin, primary units, molecular
41
diffusion, NMR
42
ACS Paragon Plus Environment
2
Page 3 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
43
1
INTRODUCTION
44
Complex coacervation results from the interaction of two oppositely charged colloids leading to a
45
liquid-liquid phase separation. The denser phase is called coacervates and the other phase is the
46
dilute phase or the equilibrium phase. There are numerous examples of complex coacervation
47
between oppositely charged biopolymers but the complex coacervation involving oppositely
48
charged proteins (heteroprotein coacervation) is rather rare. The mechanism of heteroprotein
49
complex coacervation, not completely elucidated, exhibits some specificity compared to complex
50
coacervation involving polyelectytes. Heteroprotein coacervation is observed in very narrow ranges
51
of pH, ionic strength, protein concentration and molar ratio. However, some common features for
52
heteroprotein coacervation exist even if the number of heteroprotein systems studied up to now is
53
too limited to established a general rules.
54
It is assumed that the formation of well-defined primary complex (building block) constitutes a
55
preliminary step to heteroprotein coacervation.1-4 The driving force for building block formation is
56
mainly electrostatic interactions with an entropy contribution due to counter-ions release. The
57
primary units have to be close to charge neutrality for liquid-liquid phase separation to occur and
58
consequently coacervation is only observed between the pI of the proteins involved in the
59
coacervation process. Protein size compensation is another requirement for heteroprotein
60
coacervation as two proteins of opposite charge but equal magnitude do not form coacervates if
61
they have different molecular size.5 It is also observed that the flexibility of at least one protein
62
favors heteroprotein coacervation.6-8 Protein flexibility could stabilize the building block by an
63
optimal exposition of amino acids involved in the interaction.4
64
Once formed, the primary units separate into a dense phase. Such mechanism suggests that protein
65
molar ratio in the coacervates is constant and corresponds to the protein stoichiometry of the
66
building block. This was observed for most of the systems studied up to now but the results are still
67
unclear for lactoferrin (LF) - β-lactoglobulin (β-LG) system. Under selected conditions, Anema and
68
de Kruif
9
proposed a β-LG/LF molar ratio of three while other authors proposed a molar ratio of
ACS Paragon Plus Environment
3
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 4 of 26
69
four.1,2 This difference was first explained by the origin of the proteins used for coacervation
70
experiments. Note that in these studies β-LG/LF molar ratio in the coacervates were determined
71
indirectly. By using the same protein source, we determined by direct quantification of the proteins
72
in the coacervates that the protein molar ratio in the coacervates varied from 4 to 6-8 by modifying
73
the protein molar ratio in the mixture.10 From the complexes in the dilute phase, Flanagan, et al.
74
proposed that the building block for LF - β-LG heteroprotein coacervation was a pentamer LF(β-
75
LG2)2 including two β-LG dimers bound to one LF. However, the structure of the coacervate was
76
analyzed by SANS and the results indicate that the proposed model does not exactly fit with the
77
experimental data.11 These authors proposed that the building block (or primary blocks) may adopt
78
different configurations and that some of them may associate into higher order equilibrium
79
structures. The purpose of the present study is to gain further insights into the internal structure and
80
the dynamic of the fascinating coacervates formed throughout specific interactions between
81
proteins. Here we investigate the molecular interactions and the diffusion properties of the proteins
82
in the highly concentrated LF/β-LG coacervate phase using Nuclear Magnetic Resonance (NMR),
83
Fluorescence Recovery after Photo Bleaching (FRAP) and docking simulations. The two
84
experimental technics allow access to complementary information at different time scales. Our data
85
show that one LF can bind multiple β-LG dimers with different affinities. FRAP and NMR
86
measurements support the coexistence of different molecular species in the coacervates. The
87
presence of various primary units in equilibrium constitutes a specific characteristic of BLG-LF
88
coacervates.
2
89 90
2
EXPERIMENTAL SECTION
91
2.1 Reagents and Solutions
92
Bovine β-LG, a mixture containing genetic variants A and B (pI ≈5.2), was provided by a
93
confidential industrial source. The protein powder was dispersed in deionized water (45 g/L),
94
adjusted to pH 5.2 with 1M HCl and kept at 30°C for 5 min, in order to precipitate non-native forms ACS Paragon Plus Environment
4
Page 5 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
95
of β-LG. The dispersion was centrifuged at 20,000g at room temperature for 10 min (Heraeus
96
Biofuge Primo, Thermo Scientific, Waltham, MA, USA) and the supernatant containing only native
97
β-LG was adjusted at pH 7.0 with 1 M NaOH, freeze-dried and stored at – 20 °C until use. Bovine
98
lactoferrin (LF, pI ≈8.5) (purity of 90% and iron saturation of 10 - 20 % according to technical
99
specification) was purchased from Fonterra Cooperative Group, New Zealand. MES hydrate buffer
100
was purchased from Sigma-Aldrich (St. Louis, MO, USA) and all other chemicals were from VWR
101
(Radnor, PA, USA).
102
β-LG and LF protein powders were solubilized in a 10 mM MES buffer, pH 5.5, at about 5.0 mM
103
and 0.7 mM respectively and filtered through a 0.2 µm membrane (cat. no. 4612, Pall Corporation,
104
Ann Arbor, MI, USA). The exact protein concentration was determined by absorbance at 280 nm
105
(SAFAS UV MC2, Safas, Monaco) using 0.96 L g-1 cm-1 and 1.47 L g-1 cm-1 as extinction
106
coefficients for β-LG and LF respectively. No sign of protein self-aggregation was detected in the
107
stock solutions that were transparent.
108 109
2.2 Preparation of β-LG-LF coacervates
110
The coacervates were obtained by mixing an appropriate volume of the protein stock solutions to
111
reach β-LG and LF final concentration of 0.6 mM and 0.06 mM respectively. These optimal
112
concentrations were defined in accordance with our previously work.10 The solution was prepared
113
respecting the following order of mixing: MES buffer + β-LG stock solution + LF stock solution.
114
Immediately after mixing the solution evolved in a system containing two liquid phases in
115
equilibrium: (i) a dilute phase and (ii) a dense phase dispersed in the dilute phase, called
116
coacervates. As described by Tavares, et al. 10 the coacervates were separated from the dilute phase
117
by centrifugation (Heraeus Biofuge Primo, Thermo Scientific, Waltham, MA, USA) at 20,000g for
118
10 min. Proteins in the coacervates were quantified by reverse phase chromatography as previously
119
described by Tavares, et al. 10.
120
The coacervate samples studied here were a translucent, slightly pink and viscous liquid containing ACS Paragon Plus Environment
5
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 6 of 26
121
around 150 g/kg of LF and 130 g/kg of β-LG, i.e. a β-LG/LF molar ratio about 4, which is in
122
accordance with the ratio observed by other authors.
123 124
2.3 Evaluation of β-LG and LF heterocomplex by rigid docking
125
A rigid docking experiment was conducted to determine the near-native orientation of β-LG (PDB
126
id: 1BEB) and LF (PDB id: 1BLF) using the web server pyDockWeb.12 The software creates
127
10,000 different complexes by changing the relative position of the ligand around the receptor.
128
Then a scoring energy was calculated and the complexes were ranked according. Best poses were
129
assumed to be the ones that displayed the lowest scoring energy. Three docking were performed: i)
130
LF as receptor and β-LG dimer (noted β-LG2) as ligand; ii) The best LF(β-LG2) complex obtained
131
as receptor and another β-LG dimer as ligand; iii) The best LF(β-LG2) complex against himself.
132
The 100 best complexes were clustered by their pairwise Root Mean Square deviation (RMSD) of
133
the Cα atoms with ensemble cluster tool13 embedded in Chimera software14 to reduce and
134
discriminate docking solutions. Consequently we assume that the best docking solutions were the
135
complexes in the most populated clusters. Structures were visualized with VMD.15
136 137
2.4 FRAP: analysis of β-LG binding to LF
138
FRAP analysis was used to characterize the binding between FITC (Fluorescein isothiocyanate)
139
labeled β-LG and LF. In solution where a fluorescent molecule displays a classical diffusion
140
behavior, the radius of a Gaussian bleached spot increases with time after photo-bleaching.16 FRAP
141
recovery curve can be then used to estimate the molecular diffusion constant of the fluorescent
142
molecule. In contrast, in a solution where the fluorescent molecules are strongly bound to immobile
143
obstacles, the radius of a spherical bleached spot does not increase with time.16 In this reaction
144
dominant regime, the time of diffusion of a “free” fluorescent molecule is shorter than the diffusion
145
time of a complex (with bound fluorescent probe). Thus, the large scale fluorescent probe diffusion
146
measured by FRAP is exclusively dictated by the time that the probe remains bound to the obstacle. ACS Paragon Plus Environment
6
Page 7 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
147
Therefore, FRAP recovery curve informs on the probe-obstacle dissociation rate (κoff) and on the
148
concentration of fluorescent probes bound to the obstacle (Ceq). For reaction dominant regime, the
149
FRAP recovery data can be fitted by the following equation:
150
Eq. 1.
151 152 153
Where frap(t) is the fluorescence intensity at time t and Feq is the concentration of unbound
154
fluorescent molecule.17
155
This equation only fits the experimental data if all different probe-obstacle complexes present the
156
same dissociation rate. However, we found a dependence between the lifetime of the LF(β-LG2)n
157
complexes and the number of β-LG2 bound to LF in the current work. Considering that β-LG2 binds
158
to LF in two different affinity sites,1,
159
equation:17
160
Eq. 2.
10
FRAP data were fitted using a double exponential
161 162 163
The parameters of Eq. 2 are the same as in Eq. 1 and the indices 1 and 2 refer to the complexes with
164
longer and short lifetimes, respectively.
165
Complementarily, it is also possible that one of the two β-LG2 bound to LF presents a too weak
166
affinity to be considered in the reaction dominant regime. In this case, the fraction of β-LG2
167
strongly bound to LF presents a reaction dominant regime, while the other fraction presents a
168
different regime characterized by a significant diffusion time of free β-LG2 between LF-β-LG2
169
complexes (obstacles). In this case, the following equation must be used:17
170
Eq. 3.
171
ACS Paragon Plus Environment
7
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 26
172 173
The last term in Eq. 3 describes the contribution of the β-LG2 bound to the LF site of highest
174
affinity (identical term of the Eq. 2). In contrast, the first term of the equation describes the
175
contribution of the β-LG2 fraction bound to the LF site of weakest affinity. The equation parameters
176
are the same previously described and τeff is defined as:17
177
Eq. 4.
178 179 180
Where w is the radius of the bleached spot, Deff and Df are the effective diffusion (the hindrance
181
induced by diffusion and binding) and the pure diffusion (the hindrance induced by diffusion only)
182
respectively and k
183
of hindered diffusion for β-LG2 found experimentally18 and theoretically in equivalent dense
184
systems. The best fit was determined according to the methodology described in Supporting
185
Information (Figure S1). Note that the immobility is a relative notion. If the recovery time of the
186
signal is faster than that taken by a macromolecule complex, a so called obstacle, to do a significant
187
displacement (µm scale), the last can be considered as immobile regarding the FRAP experiment.
188
This is not true for NMR because of much smaller time scales and high sensitivity to the rotation
189
motion of quite large complexes.
190
Finally, the LF(β-LG2)n complexes are not totally immobile and their diffusion is significant
191
regarding the time range of the experiment. The diffusion of these complexes themselves was
192
followed and treated as a classical but slow diffusion behavior. If a fraction of β-LG2 forms a stable
193
complex with LF in the time range of the experiment (strongly bound) and another fraction displays
194
a dynamical association (weakly bound) with LF, the Eq. 3 can be used to model the fluorescence
195
recovery curve.
196
Basically, the coacervate phase for the FRAP experiments was prepared as described above. The
on
is the association rate. The value of Df used was 15 µm2/s which is the value
ACS Paragon Plus Environment
8
Page 9 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
197
proportion of 1% of the bulk β-LG concentration was replaced by the FITC labeled β-LG. The
198
FITC-β-LG was obtained as described by Silva, et al. 19
199
The coacervate phase was placed between glass slide and a coverslip sealed with a small adhesive
200
frame (25 µL capacity). The fluorescence recovery images were obtained using an inverted CLSM
201
(Nikon, Champigny-sur-Marne, France). The observations were realised with a 40 times
202
magnification oil immersed objective at 30 µm from the coverslip. The labeled β-LG in the
203
coacervate phase was excited with 50 mW sapphire laser system at a wavelength of 488 nm and
204
detected from 500 to 530 nm. The fluorescence before photobleaching was recorded using 0.1% of
205
the maximum laser intensity. The photobleaching was realized using 100 % of the laser intensity in
206
an 85 µm² spot and the recovery of fluorescence was followed during 60 min. Obtained images
207
were treated using ImageJ free software.
208 209 210
2.5 NMR: Estimation of hydrodynamic radius and relative abundance of the different β-LG/LF complexes
211
Solid-state NMR experiments were performed on a Bruker Avance III 600 SB (14T) operating at
212
Larmor frequencies of 600.1 MHz for 1H using a 4 mm MAS double resonance probehead. 1H
213
NMR spectra were acquired under static or slow Magic Angle Spinning (MAS at 500 Hz)
214
conditions using a single pulse experiment (t90 1H = 3.6 µs). The recycle delays used to record
215
proton spectra were set to 1 s which was long enough to ensure quantitatively. Moreover, 1H spectra
216
of the sample remained unchanged after 24 h, indicating that the analyzed suspension was stable
217
under MAS as well as in static conditions. Longitudinal (T1) and transverse (T2) relaxation times
218
were measured using a saturation/recovery and spin echo pulse sequences, respectively. 1H spectra
219
were referenced according to Tetramethylsilane (TMS). Spectra deconvolutions were performed
220
using the dmfit softwere as detailed in Supporting Information. To check the repeatability of the
221
protocol, relaxation time has been measured in three different samples.
ACS Paragon Plus Environment
9
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 10 of 26
222
The hydrodynamic radius (Rh) of β-LG2, LF and the formed complexes was estimated from their
223
rotation correlation time (τc), which is strongly correlated to the 1H translation relaxation times, T1
224
and T2. In diluted solution, τc is related to Rh of the molecule as described below:20
225
Eq. 5.
226 227 228
Where η is the solvent viscosity, kb is the Boltzmann constant and T is the temperature.
229
In dense solution, the hydrodynamic interactions between particles are affected by the solvent
230
viscosity (η), the determination of which is not trivial in particular for polydisperse systems.21 The
231
value of η increases with the molecular crowding and the size of the diffusive particles. Theoretical
232
studies showed that the molecular motion of small particles (~1 to 7 nm of radius) displays a quasi-
233
linear relationship with Rh at short time scales (nano to miliseconds), in protein density range
234
equivalent to the ones of the present study (250 to 350 g/L).22,23 Thus, even if it is not possible to
235
exactly determine η (since the composition of the medium is unknown) the quasi-linear relationship
236
between the size of the molecule and τc can be used to roughly estimate a relative Rh for the
237
different species under study. More details on the Rh estimation from relation times are given in
238
Supplementary Information.
239 240
Determination of the relative abundance of the different β -LG/LF complexes from the NMR
241
1
242
The peak width in NMR is very sensitive to the dynamics of the molecular species in solution. If the
243
association/dissociation rate of the complexes is slow regarding NMR time scales (nano to
244
milliseconds), peaks with different widths are observed on a 1D 1H spectrum: complexes with slow
245
dynamics produce broad peaks while their primary units having fast dynamics exhibit narrow peaks.
246
Based on this, the relative abundance of the complexes was directly estimated by measuring the
H spectrum.
ACS Paragon Plus Environment
10
Page 11 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
247
Langmuir
signal intensity of each peak in a 1D 1H spectrum.
248 249
3
RESULTS
250
3.1 β-LG and LF complexes: rigid Docking
251
The association of one β-LG dimer (β-LG2) and one LF has been evaluated by rigid-body docking
252
using PyDockWeb server. The 100 lowest-energy complexes structures were clustered by pairwise
253
RMSD to reduce docking solutions. Nineteen clusters were obtained and a representative structure
254
by cluster was extracted for further analyses. All of them showed the same LF region (called S for
255
single) interacting strongly with β-LG dimer (Figure 1A). The S site of LF exhibited a very high
256
density of positively charged residues (inset of Figure 1A) suggesting electrostatic interactions as
257
the main driving forces of protein association. Different orientations of β-LG2 were observed
258
supporting various interaction sites on β-LG2 surface. The associated LFβ-LG2 complex was further
259
used as receptor to simulate a second interacting site on LF. The 100 best docking solutions were
260
clustered and highlighted various interaction sites. Two of the major interaction sites were identified
261
in LF (called M for Multiple) representing 48 and 31 % among the best poses (Figure 1B). These M
262
sites are not identified in the first docking experiment, suggesting the involvement of lower binding
263
forces. Furthermore, M sites are localized on the surface in the hinge between the two lobes of LF.
264
This suggests that LF can bind at least one β-LG2 in addition to anther β-LG2 already linked to S
265
site of LF.
ACS Paragon Plus Environment
11
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 26
266 267
Figure 1. Molecular docking between β-LG2 and LF (A), and between β-LG2 and already formed
268
LFβ-LG2 complex (B). (A) Structure of the best complex formed by β-LG2 in red and LF in blue.
269
Other docking solutions of β-LG2 are represented in translucent grey. β-LG2 binds a specific site S,
270
in all docking solutions. The inset shows that the S site on LF exhibits a huge density of positively
271
charged residues (blue) in contrast to the other parts of the protein with positively (blue) and
272
negatively (red) charged residues homogeneously distributed. (B) Illustration of the secondary sites
273
on LF (called site M) where a second β-LG2 binds (with indicated percent) to the LFβ-LG2
274
complex.
275 276
3.2 β-LG binding within the coacervate phase: FRAP analysis
277
Figure 2A shows a bleached spot profile, immediately (0.5 s) and one hour after bleaching. Figure
278
2B shows the corresponding FITC labeled β-LG fluorescence recovery curve. The radius of the
279
bleached spot profile displays only a small increase one hour after bleaching indicating that the
280
diffusion of β-LG2 within the coacervates is dominated by the binding of β-LG2 to immobile
281
obstacles (reaction regime, see experimental section for details). Only 70% of initial β-LG
282
fluorescence intensity was recovered one hour after bleaching. This is a remarkable slow recovery ACS Paragon Plus Environment
12
Page 13 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
283
even for a system as dense as this one17, 20 indicating that (i) β-LG binds to the obstacles for a long
284
time and (ii) the obstacles are large enough to consider their diffusion negligible after one hour.
285
Since the bleached spot analysis indicates that the system is reaction dominated, the recovery curve
286
has been fitted assuming a model in which the binding of β-LG2 to LF dominate the fluorescence
287
recovery. A model in which one single dissociation time for all complexes is considered did not
288
correctly fit the experimental data (Figure 2B, red line). Proper fits were obtained with models
289
considering either two different dissociation times for the complexes (Figure 2B, green continuous
290
line) or both the diffusion of β-LG and its binding to complexes (Figure 2B, green dotted line). The
291
fitting parameters of the last two cited models are displayed in Table 1.
292
293 Figure 2. FRAP analysis of the FITC labeled β-LG within the coacervate phase. (A): Time evolution of the bleached spot profile from a FRAP experiment using an initial bleached radius of 5.2 µm. (B): Typical FRAP recovery curve (blue) and the fits using the models described in materials and methods section. The red line represents the best fit using the model considering the diffusion of β-LG2 dominated by binding and that all complexes display the same lifetime. The green continuous curve represents the best fit using a model considering the diffusion of β-LG2 dominated by its binding to complexes which display two different lifetimes. The green dotted curve represents the case where both, β-LG2 diffusion and its binding to complexes were considered for the fluorescence recovery. 294 ACS Paragon Plus Environment
13
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 14 of 26
Table 1. FRAP fitting parameters
Feq (%)
Reaction dominant 6
Reaction/Diffusion 70*
Ceq1 (%)
45
70*
Ceq2 (%)
49
30
k1 off (s-1)
1.0 x 10
-3
8.6 × 10-3
k2 off (s-1)
1.2 x 10-4
5.2 × 10-4
k1 on (s-1)
-
30
295
15 Df (µm²/s) * In the reaction/diffusion model, Feq and Ceq1 (see equation 3 in experimental section) cannot be
296
distinguished; the value of 70% corresponds to both Feq and Ceq1.
297 298
The dissociation rates k1off and k2off range from 1 to 8 ms-1 and 0.1 to 0.5 ms-1 for the pure reaction
299
and reaction/diffusion models, respectively (Table 1). The dissociation times are slightly lower in
300
the pure reaction model than in the reaction/diffusion model. However, in both models, the first
301
dissociation rate (k1off) was about one order of magnitude higher than the second constant (k2off). In
302
contrast, the relative number of β-LG molecules (Ceq) associated to k1off and k2off was quite
303
different. The number of molecules associated to both rates was around 50/50 in the reaction
304
dominant model. In contrast, more molecules (70%) were associated to the first dissociation rate
305
compared to the second one in the reaction/diffusion model.
306 307 308
3.3 Rotation diffusion coefficients show the presence of three different complexes: NMR analysis
309
Besides the water signal at 4.9 ppm, three signals were distinguished in the 1H spectrum of the
310
coacervate phase (Figure 3A): a large signal with a full width at half maximum (fwhm) of about 50
311
ppm, a second signal with 20 ppm of fwhm and a narrow signal of 2-4 ppm of fwhm (see
312
Supporting Information for details). Protons with the largest fwhm also present the slowest
313
dynamic. The narrow signal displays the typical pattern of water and the others signals are assigned
ACS Paragon Plus Environment
14
Page 15 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
314
to protons of proteins. The 1H 20 ppm and the 1H 50 ppm corresponds to complexes displaying a
315
relative fast and slow dynamics respectively. Noteworthy, this was further confirmed on the 2D 1H-
316
13
317
dynamics appeared. The CP spectrum displays only the signal of the largest Gaussian with a fwhm
318
of 50 ppm (green dotted lines in Figure 3A). A focus at the center of the 1H spectrum (Figure 3C)
319
shows individual peaks assigned to the amino acids of the proteins composing the second signal and
320
the underneath spectrum of the slowest species described above. Fig. 3D shows that the signal of
321
the slowest species is quite weak after 20 µs using an echo experiment (that measures T2). In
322
contrast, the intensity of the second signal barely changed confirming that their relaxation times,
323
and consequently their dynamics, are extremely different. The deconvolution of the 1H spectrum
324
gave the fraction of protons assigned to the slowest species (Supplementary Information). The peak
325
of the slowest species (largest signal) represents about 52% of the total 1H NMR signal intensity of
326
the proteins, against 48% for the second signal composed of two species (see below): 16% and 32%
327
for the fastest and the moderately slow species, respectively (Table 2).
328
For comparison, Figure 3E and F show the spectra of the solutions of β-LG (Figure 3E) and LF
329
(Figure 3F) at 150 g/L, under the same conditions of pH and temperature. Similar results are
330
obtained when denser solutions (~300 g/L) of β-LG or LF were analyzed.
C cross-polarization (CP) spectrum (Figure 3B) on which only the protons displaying a very slow
ACS Paragon Plus Environment
15
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 16 of 26
331 332
Figure 3. NMR signal Assignments. (A) Three main signals detected in 1H spectrum of the
333
coacervate phase: the largest signal is assigned to the association of several LF and β-LG2
334
molecules i.e. (LFβ-LG2)n, the second signal is assigned to β-LG /β-LG2 and LF(β-LG2)2 and the
335
third signal is assigned to water molecules. (B) 2D 1H-13C cross polarization spectrum of the
336
coacervate phase. In the spectrum only the protons having a very slow dynamic display a reasonable
337
intensity. (C) Zoom at the center of the spectrum A focusing on the peaks emanating from β-LG2
338
and LF(β-LG2)2. (D) 1H spectrum using echo filter. In the spectrum between 5 and 20 µs the signal
339
of the largest peak is completely eliminated without affecting significantly the intensity of the other
340
peaks (the water and the second peak corresponding to β-LG2 and LF(β-LG2)2. (E) and (F), 1H
341
spectra of dense β-LG and LF solutions respectively.
342 343
The measurement of the spin relaxation times T1 and T2 gives access to quantitative data about the
344
protein dynamics (notably the protein rotational correlation time as detailed in Supplementary
ACS Paragon Plus Environment
16
Page 17 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
345
Information). Figure 4 shows the spectra at different delays (from 0 to 3 s) used to measure T1.
346
After deconvolution, two mean protein signals with different dynamics were obtained. The
347
measurement of the T1 of each signal can be determined independently (see experimental section
348
and Figures S2 and S3). For the detected signals, the time evolution echo can be fitted with either
349
mono-exponential (largest peak) or double-exponential (second peak) curves. This indicates that
350
relaxation times are dominated by the global dynamics of the proteins. Based on hydrodynamic
351
radius calculation (Table 2), the largest signal is assigned to complexes composed of several LF and
352
β-LG2 noted (LFβ-LG2)n. From calculated Rh, molecular species associated to the second signal
353
were assigned to β-LG /β-LG2 and LF(β-LG2)2, respectively.
354 355
356 Figure 4. Zoom-in view of NMR signals (insert) for assignment to different complexes in the coacervates. Spectra corresponding to the different delays used to calculate T1 (delays from 0 to 3 s): Green delimited area: large complexes; red delimited area: small molecular entities.
ACS Paragon Plus Environment
17
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 18 of 26
Table 2. Abundance (w/w) and dynamics of the complexes in the coacervate phase % Protons
T1 /T2
% β -LG*
Rh (nm)**
β-LG/β /β-LG /β 2
16
2
≈ 33
2
LF(β β-LG2)2
32
17
≈ 33
7
(LFβ β-LG2)n
52
200
≈ 33
30-60
357
* Relative proportion of β-LG molecules in each complex, calculated on the basis of NMR data.
358
** Calculated using the values of T1/T2 ratio (see experimental section and Supplementary
359
Information).
360 361
4
DISCUSSION
362
4.1 Primary units of the coacervates
363
It has been postulated that LF-β-LG coacervates result from the association of heterocomplexes
364
formed by LF and β-LG2.1, 10 Docking simulations showed that each LF molecule is able to bind 2
365
or more β-LG2. The first β-LG2 always bind to the same site of higher affinity on LF surface (site S,
366
Figure 1B). In contrast, the second β-LG2 can bind on different sites of lower affinity, called M
367
sites, located all around LF surface. The theoretical evidence of the existence of two sites of
368
different affinity for β-LG (sites S and M) on each LF confirms the results obtained previously by
369
isothermal calorimetry (ITC)10 and electrostatic modeling.2 The coexistence of these
370
heterocomplexes was already suggested by Flanagan, et al. 2. Their relative abundance depends on
371
β-LG/LF initial molar ratio. In support to docking simulation, FRAP experiments show that β-LG
372
bind to immobile obstacles with two quite different affinities. The immobile obstacles could be
373
either an individual LF monomer or an already formed complex. The fit of FRAP data reveals
374
dissociation rates of about 1 to 8 ms-1 for M sites and 0.1 to 0.5 ms-1 for the S site. Both these values ACS Paragon Plus Environment
18
Page 19 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
375
correspond to a relative weaker binding strength compared to the ones described for most biological
376
complexes displaying a specific binding (dissociation rate of about 10-3 ms-1 for protein receptors).17
377
Given the strong affinity of LF S-site for β-LG2, the probability of having free LF in the coacervates
378
should be weak under molar excess of β-LG compared to LF. NMR data indicate that β-LG
379
distributes equally to the various species in the coacervate i.e. the free β-LG and the two
380
heterocomplexes LF(β-LG2)2 and (LFβ-LG2 )n (see below). From HPLC quantification, 7.2 mM β-
381
LG were found in the coacervates meaning 2.4 mM in each of the above species. Consequently, the
382
LF concentrations involved in LF(β-LG2)2 and (LFβ-LG2)n heterocomplexes are expected to be 0.6
383
mM and 1.2 mM, respectively. This match perfectly with the quantified LF value of 1.87 mM found
384
in the coacervate phase in which, consequently, less than 5% of LF exists as free entity. Hence,
385
based on protein quantification and ITC experiments10, we suggest that under these conditions, LF
386
molecule cannot bind, in significant proportion, more than two β-LG2 simultaneously.
387 388 389
4.2 Identification of the structures and quantification of the main species in the coacervates
390
Based on our experiments, three types of molecular entities with specific dynamics and Rh seem to
391
be present in LF-β-LG coacervates. The smaller one is assigned to β-LG2 and β-LG monomers (Rh
392
= 2 nm). Considering a dissociation constant Kd of 5.5×10-4 M at working pH of 5.5
393
LG2/β-LG molar ratio is roughly 60/40 for the non-complexed β-LG in the coacervates (2.4 mM).
394
The other entities with Rh of 7 nm and ≈30-60 nm are assigned to heterocomplexes involving LF
395
and β-LG2. We suggest that the complex with 7 nm determined by NMR corresponds to LF(β-
396
LG2)2 while complexes with Rh ranging from ≈ 30 to 60 nm results from the association of several
397
LFβ-LG2 units to form (LFβ-LG2)n, the “skeleton of the coacervates”. In this skeleton, we suggest
398
that the proportion of LF molecules having more than two branched β-LG2 is limited due to steric
399
hindrance leading to the specific and oriented growth of n×LFβ-LG2 into (LFβ-LG2)n
ACS Paragon Plus Environment
10
, the β-
19
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 20 of 26
400
heterocomplexes. An association of LF(β-LG2)2 pentameter at the ends of this skeleton structure
401
hinders its further growth leading to a particle definite size. The presence of free LFβ-LG2 which
402
has a Rh close to the one of LF(β-LG2)2 in the coacervates is not excluded. But for a sake of
403
simplification, we assume from NMR data that the main entities in the coacervates are β-LG2, LF(β-
404
LG2)2 and (LFβ-LG2)n with a relative proportions, based on proton assignment, of 17%, 33% and
405
50% respectively (Table 2). Since the number of protons of β-LG represents ∼50% and ∼33% of the
406
total number of protons in LF(β-LG2)2 and (LFβ-LG2)n respectively, we deduced that β-LG
407
molecules in the coacervates distributes almost equally between the three entities. Based on the
408
protein composition in the coacervates determined by NMR, a rapid calculation of the β-LG/LF
409
molar ratio gives ∼4, a value in agreement with the ratio found by protein quantification previously
410
reported.1,10 Combining our results, we propose a schematic molecular composition of the β-LG-LF
411
coacervate phase (Figure 5). This phase exhibits heterogeneous composition with a subtle mix of
412
free β-LG, heterooligomers and heterocomplexes. The number of primary units that forms the
413
heterocomplexes has been chosen considering a Gaussian distribution. They are diluted in a “sea”
414
of β-LG2 and LF(β-LG2)2 complexes. Under the experimental conditions, the number of β-LG2 is
415
close to twice the number of LF(β-LG2)2.
416 417
Figure 5. Two-dimensional representation of the β-LG-LF coacervate phase composition. LF: blue ACS Paragon Plus Environment
20
Page 21 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
418
full ellipses; β-LG monomers or dimers: red full spheres. Dark circles indicate large
419
heterocomplexes containing several LF and β-LG2 (LFβ-LG2)n. Red circles indicate LF(β-LG2)2
420
complexes. The ratio of each species is derived from NMR data as described in the text.
421 422
4.3 Specific thermodynamic equilibrium governs the stability of the coacervates
423
The LF(β-LG2)2 pentamer was reported to be the primary unit of the LF-β-LG coacervate.11 This
424
suggests that LF(β-LG2)2 entities are very abundant and stable in solution. Consequently, the
425
conditions of formation of LF-β-LG coacervates should be quite resistant to variation of both total
426
protein concentration and β-LG/LF molar ratio. Experimentally, we observed that random
427
aggregates instead of coacervates were formed in mixtures containing β-LG/LF molar ratio as high
428
as 20.10 Increasing β-LG concentration in LF solution should drive the binding equilibrium toward
429
the saturation of the LF binding sites resulting in larger amount of LF(β-LG2)2 and/or LF with more
430
than two bound β-LG2. These multi-branched LF would lead to non-specific associations into
431
random aggregates.
432
The results presented here give an explanation to previous indirect assumptions. We propose that
433
the coacervates result from the coexistence of three metastable entities in dynamic equilibrium: β-
434
LG2, LF(β-LG2)2 and larger complexes (LFβ-LG2)n. Structures with larger size were not detected in
435
the dilute phase according to Flanagan et al.2 and confirmed here by DLS measurements (not
436
shown). Hence, unlike the small molecular species, (LFβ-LG2)n is probably in equilibrium with
437
other species only in the coacervate phase. Any subtle changes of these equilibria by modifying the
438
physicochemical conditions (protein concentration, β-LG/LF molar ratio, ionic strength and pH)
439
could change the relative abundance of each entity. This could explain the reported evolution of the
440
β-LG/LF molar ratio in the coacervation domain and also the transition from coacervates to
441
aggregates for larger changes.10 It could also explain the extreme sensitivity of LF-β-LG
442
coacervation by changing ionic strength and pH when compared to similar liquid-liquid systems
ACS Paragon Plus Environment
21
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 22 of 26
443
involving large colloids and polyelectrolytes. Ionic strength and pH conditions affect the
444
equilibrium between electrostatic forces responsible for the long/short range attraction/repulsion in
445
most liquid-liquid systems. The properties of LF-β-LG coacervates are dictated by the balance of
446
the interaction between entities and their relative abundance. Any changes in the physico-chemical
447
parameters have a double impact on the present system: it modifies the self-repulsion/attraction
448
properties of each complex (as for colloids and polyelectrolytes) but also their relative abundance.
449
This explains why the present system is much more sensitive to changes in physicochemical
450
conditions than the ones composed by more stable units (colloids or polyelectrolytes). Complexes
451
with constant stoichiometry were generally reported as primary units for coacervation between
452
oppositely charged polyelectrolytes. Here, we evidence the co-existence of complexes with various
453
stoichiometries in dynamic equilibrium forming the inner structure of heteroprotein coacervates.
454
We therefore confirm the existence of a dynamic equilibrium into LF-β-LG coacervates suggested
455
by Dubin’s group from rheology and neutron scattering experiments.11
456 457
To conclude, this study gives new insights on the structure and dynamic of heteroprotein
458
coacervates. The stability of the heteroprotein coacervates is governed by the presence of non-
459
random nano-complexes in fast equilibrium with smaller entities. The structure of the nano-
460
complexes and their dynamic prevents the formation of large, random aggregates at mesoscale and
461
is responsible of the liquid behavior of the dense phase at macroscopic scale. In the context of drug
462
delivery systems, these data indicate that the key factor governing the formation of the delivery
463
protein vehicle are the association rates of the proteins involved rather than the specific properties
464
of the complexes formed. Thus, the scientific effort to identify physico-chemical conditions for
465
optimizing the formation of delivery protein vehicle should target on a detailed study of protein
466
association dynamics. Also, further studies are still needed to claim generality of our finding on LF-
467
β-LG to other heteroprotein systems.
468
ACS Paragon Plus Environment
22
Page 23 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
469
Supporting Information
470
The Supporting Information is available free of charge at http://pubs.acs.org. Modeling of FRAP
471
data; Estimation of the hydrodynamic radii from NMR experiments: i- Measure of relaxation times,
472
ii- spectra simulations.
473 474 475
5
REFERENCES
476
(1)
477
Schmitt, C. Heteroprotein complex coacervation: bovine beta-lactoglobulin and lactoferrin.
478
Langmuir 2013, 29, 15614-23.
479
(2)
480
Bovetto, L.; Schmitt, C. Complex equilibria, speciation, and heteroprotein coacervation of
481
lactoferrin and beta-lactoglobulin. Langmuir 2015, 31, 1776-83.
482
(3)
483
2016, 55, 89-99.
484
(4)
485
R.; Schwalbe, H.; Croguennec, T.; Bouhallab, S., et al. Investigation at Residue Level of the Early
486
Steps during the Assembly of Two Proteins into Supramolecular Objects. Biomacromolecules 2011,
487
12, 2200-2210.
488
(5)
489
drive spontaneous self-assembly of oppositely charged globular proteins into microspheres. J. Phys.
490
Chem. B 2010, 114, 4138-44.
491
(6)
492
supramolecular structures resulting from alpha-lactalbumin-lysozyme interaction. Biochemistry
493
2007, 46, 1248-55.
Yan, Y.; Kizilay, E.; Seeman, D.; Flanagan, S.; Dubin, P. L.; Bovetto, L.; Donato, L.;
Flanagan, S. E.; Malanowski, A. J.; Kizilay, E.; Seeman, D.; Dubin, P. L.; Donato-Capel, L.;
Delboni, L. A.; da Silva, F. L. B. On the complexation of whey proteins. Food Hydrocolloid
Salvatore, D.; Duraffourg, N.; Favier, A.; Persson, B. A.; Lund, M.; Delage, M. M.; Silvers,
Desfougeres, Y.; Croguennec, T.; Lechevalier, V.; Bouhallab, S.; Nau, F. Charge and size
Nigen, M.; Croguennec, T.; Renard, D.; Bouhallab, S. Temperature affects the
ACS Paragon Plus Environment
23
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 24 of 26
494
(7)
Anema, S. G.; de Kruif, C. G. Coacervates of lysozyme and beta-casein. J. Colloid Interface
495
Sci. 2013, 398, 255-61.
496
(8)
497
Milk after the Binding of Lactoferrin or Lysozyme. J Agr Food Chem 2013, 61, 7142-7149.
498
(9)
499
lactoglobulin. J. Colloid Interface Sci. 2014, 430, 214-20.
500
(10)
501
coacervation between lactoferrin and the two isoforms of β-lactoglobulin. Food Hydrocolloid 2015,
502
48, 238-247.
503
(11)
504
Schmitt, C. Structure of bovine beta-lactoglobulin-lactoferrin coacervates. Soft Matter 2014, 10,
505
7262-8.
506
(12)
507
protein-protein docking using electrostatics and desolvation scoring. Bioinformatics (Oxford,
508
England) 2013, 29, 1698-9.
509
(13)
510
ensemble of NMR-derived protein structures into conformationally related subfamilies. Protein Eng
511
1996, 9, 1063-1065.
512
(14)
513
C.; Ferrin, T. E. UCSF Chimera--a visualization system for exploratory research and analysis. J.
514
Comput. Chem. 2004, 25, 1605-12.
515
(15)
516
1996, 14, 33-8, 27-8.
517
(16)
518
of proteins in complex with inositol lipids serves to coordinate free movement with spatial
519
information. J. Cell Biol. 2009, 184, 297-308.
Anema, S. G.; de Kruif, C. G. Protein Composition of Different Sized Casein Micelles in
Anema, S. G.; de Kruif, C. G. Complex coacervates of lactotransferrin and beta-
Tavares, G. M.; Croguennec, T.; Hamon, P.; Carvalho, A. F.; Bouhallab, S. Selective
Kizilay, E.; Seeman, D.; Yan, Y.; Du, X.; Dubin, P. L.; Donato-Capel, L.; Bovetto, L.;
Jimenez-Garcia, B.; Pons, C.; Fernandez-Recio, J. pyDockWEB: a web server for rigid-body
Kelley, L. A.; Gardner, S. P.; Sutcliffe, M. J. An automated approach for clustering an
Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E.
Humphrey, W.; Dalke, A.; Schulten, K. VMD: visual molecular dynamics. J. Mol. Graph.
Hammond, G. R.; Sim, Y.; Lagnado, L.; Irvine, R. F. Reversible binding and rapid diffusion
ACS Paragon Plus Environment
24
Page 25 of 26
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Langmuir
520
(17)
Sprague, B. L.; McNally, J. G. FRAP analysis of binding: proper and fitting. Trends Cell
521
Biol. 2005, 15, 84-91.
522
(18) Peixoto, P. D.; Bouchoux, A.; Huet, S.; Madec, M. N.; Thomas, D.; Floury, J.; Gésan-Guiziou,
523
G. Diffusion and Partitioning of Macromolecules in Casein Microgels: Evidence for Size-
524
Dependent Attractive Interactions in a Dense Protein System. Langmuir 2015, 31(5), 1755-1765.
525
(19)
526
the influence of the charge and shape of solutes on diffusion. J. Dairy Sci. 2013, 96, 6186-98.
527
(20) Rule, G. S.; Hitchens, T. K. Fundamentals of protein NMR spectroscopy. Focus on structural
528
biology. 2006, vol 5, Springer New Delhi, India.
529
(21)
530
NMR 1993, 3, 335-348.
531
(22)
532
macromolecular motion. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 18457-18462.
533
(23) Cai, L. H.; Panyukov, S.; Rubinstein, M. Mobility of nonsticky nanoparticles in polymer
534
liquids. Macromolecules, 2011, 44(19), 7853-7863.
Silva, J. V.; Peixoto, P. D.; Lortal, S.; Floury, J. Transport phenomena in a model cheese:
Boulat, B.; Bodenhausen, G. Measurement of proton relaxation rates in proteins. J. Biomol.
Ando, T.; Skolnick, J. Crowding and hydrodynamic interactions likely dominate in vivo
535
ACS Paragon Plus Environment
25
Langmuir
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 26 of 26
Peixoto et al. Abstract graphic
ACS Paragon Plus Environment
