pH-Induced Structural Transitions during ... - ACS Publications

2 Dec 2004 - and Coacervation of β-Lactoglobulin and Acacia Gum ... 2 avenue de la Foreˆt-de-Haye, BP 172, 54505 Vandoeuvre-le`s-Nancy, France, and ...
0 downloads 0 Views 362KB Size
386

Langmuir 2005, 21, 386-394

pH-Induced Structural Transitions during Complexation and Coacervation of β-Lactoglobulin and Acacia Gum Ghozlene Mekhloufi,† Christian Sanchez,*,† Denis Renard,‡ Sandrine Guillemin,† and Joe¨l Hardy† Laboratoire de Physico-chimie et Ge´ nie Alimentaires, ENSAIA-INPL, 2 avenue de la Foreˆ t-de-Haye, BP 172, 54505 Vandoeuvre-le` s-Nancy, France, and INRA, Centre de Recherches Agro-Alimentaires, Unite´ de Physico-chimie des Macromolecules, BP 71627, 44316 Nantes cedex 03, France Received May 28, 2004. In Final Form: September 28, 2004 pH-induced structural changes during complex coacervation between β-lactoglobulin (BLG) and Acacia gum (AG) in aqueous solutions were determined by coupling slow in situ acidification of BLG/AG mixed dispersions and different experimental methods. The combined signal evolution of dynamic light scattering at 90° scattering angle (I90), electrophoretic mobility, turbidimetry (τ), circular dichroism, and phase contrast microscopy allowed the distinction of critical structural transitions and the definition of their corresponding pH. The formation of soluble BLG/AG complexes was initiated at pHsc (4.90), since I90 and τ significantly increased from the baseline. In parallel or just following complexation, a conformational change of BLG was detected at pHpct (4.8). An increase in positive charge density of BLG induced complex aggregation at pHca (4.7). More efficient charge neutralization of aggregated complexes, especially through the lowering of the number of AG negative charges, promoted initiation of phase separation at pHpsi (4.4). Mixed dispersions became unstable and phase separation occurred at pHps (4.2). The phase separation of mixed dispersions was suggested by the maximum value of scattered light, by an important acceleration of the dispersion turbidity, by a strong increase of hydrodynamic radii, and by the first appearance of light fluctuations as observed by phase contrast microscopy. At the microscopic level, the first coacervates were observed at pHcoa (4.0), near the pH of the maximum of turbidity. It was also noticed that, from the onset of interactions between biopolymers, the pH decrease led to (i) a gradual homogenization of particle size in the mixed dispersion as suggested by the decrease of dispersion polydispersity and (ii) conformational transitions of the protein (a loss of R-helix structure at pHpct and a gain in protein secondary structure near pHcoa, probably involving β-sheet components).

Introduction Biopolymer mixtures in aqueous solution are generally characterized by a thermodynamic instability that results macroscopically in a phase separation. Complex coacervation (i.e., a liquid-liquid phase separation) is a particular case of associative phase separation, which is induced by electrostatic interactions between oppositely charged polymers. Great interest was shown in complex coacervation because of its implication in many biological processes like self-assembly of biological macromolecules1-3 and its use in many industrial applications such as microencapsulation,4,5 protein separation6 and purification,7 and complex food ingredients.8,9 Many studies were related to parameters controlling phase separation in * To whom correspondence should be addressed. E-mail: [email protected]. † Laboratoire de Physico-chimie et Ge ´ nie Alimentaires, ENSAIAINPL. ‡ INRA, Centre de Recherches Agro-Alimentaires, Unite ´ de Physico-chimie des Macromolecules. (1) Tribet, C.; Porcar, I.; Bonnefont, P. A.; Audebert, R. J. Phys. Chem. B 1998, 102, 1327-1333. (2) Tsuchida, E. J. Mol. Sci.-Pure Appl. Chem. 1994, A31, 1-15. (3) Kaibara, K.; Watanabe, T.; Miyakawa, K. Biopolymers 2000, 53, 369-379. (4) Burgess, D. J. In Macromolecular Complexes in Chemistry and Biology; Dubin, P. L., Bock, J., Davis, R., Schulz, D. N., Thies, C., Eds.; Springer-Verlag: Berlin, 1994; p 281. (5) Burgess, D. J.; Ponsart, S. J. Microencapsulation 1998, 15 (5), 569-579. (6) Strege, M. A.; Dubin, P. L.; West, J. S.; Flinta, C. D. In Protein purification: from molecular mechanisms to large scale processes; Ladisch, M., Wilson, R. C., Painton, C. C., Builder, S. E., Eds.; ACS Symposium Series; American Chemical Society: Washington, DC, 1990; Vol. 427, p 66.

protein-polyelectrolyte mixtures. The most important are pH, ionic strength, type of ions, protein-to-polyelectrolyte ratio, total polymer concentration, size, shape, charge density, and flexibility of macromolecules.10-14 Investigations of the structural behavior of polyelectrolyte-polyelectrolyte dispersions permitted determination of the hypothetical kinetic mechanism of complex coacervation. The Veis-Aranyi theory15 described the complex coacervation in gelatin systems as a two-step phase separation. Charged molecules interact first through electrostatic interactions and then aggregate. The neoformed aggregates (identified later as soluble intrapolymeric complexes) slowly rearrange in time to form droplets called coacervate. This theory was subsequently confirmed and completed by an intermediate step. The primary soluble complexes interact to form electrostatically neutral interpolymeric complexes. These insoluble complexes ultimately form coacervates, which coarsen with time and sediment to form the so-called coacervated phase.16,17 (7) Kabanov, V. A.; Evdakov, V. P.; Mustafaev, M. I.; Antipina, A. D. Mol. Biol. 1977, 11, 582-597. (8) Sanchez, C.; Paquin, P. In Food Proteins and Their Applications; Damodaran, S., Paraf, A., Eds.; Marcel Dekker: New York, 1997; p 503. (9) Tolstoguzov, V. Int. Rev. Cytol. 2000, 192, 3-31. (10) Bungenberg de Jong, H. G. In Colloid Science; Kruyt, H. G., Eds.; Elsevier: Amsterdam, 1949; Vol. 2, p 232. (11) Burgess, D. J. J. Colloid Interface Sci. 1990, 140, 227-238. (12) Xia, J.; Dubin, P. L. In Macromolecular Complexes in Chemistry and Biology; Dubin, P. L., Bock, J., Davis, R., Schulz, D. N., Thies, C., Eds.; Springer-Verlag: Berlin, 1994; p 247. (13) Schmitt, C.; Sanchez, C.; Desobry-Banon, S.; Hardy, J. Crit. Rev. Food Sci. Nutr. 1998, 38 (8), 689-753. (14) Doublier, J. L.; Garnier, C.; Renard, D.; Sanchez, C. Curr. Opin. Colloid Interface Sci. 2000, 5, 202-214. (15) Veis, A.; Aranyi, C. J. Phys. Chem. 1960, 64, 1203-1210.

10.1021/la0486786 CCC: $30.25 © 2005 American Chemical Society Published on Web 12/02/2004

pH-Induced Structural Changes

Structural and morphological transitions were highlighted in polymer-polymer and protein-polyelectrolyte dispersions by using a variety of experimental techniques such as turbidimetric titration,18-22,24,25 quasi-elastic and static light scattering,16,18,19,22-25 and electrophoretic light scattering,16,22 upon changing gradually the pH. Two remarkable pH values, pHc and pHφ, were identified and correspond to structural and morphological changes.19,20,24 Determined by a slight increase of scattered light intensity in dispersion, the critical pH, pHc, is related to the primary complexation of macromolecules and the formation of intrapolymeric complexes. The pHc can be considered as a phase transition on the molecular scale. The pHφ is determined by a significant increase of the dispersion turbidity, indicating, according to most of authors, the first step of phase separation at the microscopic level, leading to the coacervate droplet formation. In addition to the two pH values already defined, Kaibara et al.24 established on bovine serum albumin (BSA)-poly(dimethyldiallylammonium chloride) (PDADMAC) dispersions other characteristic pH transitions using light scattering and spectrophotometric measurements. The formation of soluble primary complexes was initiated at pHc and completed at “pH′crit”. An afterward increase in scattering intensity at “pHpre” may arise from the assembly of quasi-neutralized primary complexes. The pH region between pH′crit and pHpre was identified as the region of stable intrapolymer complexes as spectrophotometric data and pH values remained constant once the titration of the BSA-PDADMAC dispersion was interrupted. At pHφ, a maximum in scattering intensity was concomitant with both the appearance of turbidity and the first microscopic observation of coacervate droplets. Contrary to the previously mentioned definition, the pHφ corresponds in this case to a late stage of phase separation since droplets appeared. Coacervates displayed morphological changes at “pHmorph”, observed by phase contrast microscopy, followed by the transformation to solid or flocculant substances at “pHprecip”. Except for this study, all analyses of the kinetic mechanism of phase separation on a great number of polymer-polymer and protein-polymer systems, and on a more limited number of protein-polysaccharide systems,25,26,27,28 revealed that pHc and pHφ were the most important pH of structural transitions. This description seems to be incomplete with regard to the several transitions described above in the proteinpolymer phase separation process, especially considering that soluble intrapolymeric complexes have to interact in order to form, more or less, neutral aggregated (inter(16) Xia, J.; Dubin, P. L.; Kim, Y.; Muhoberac, B. B.; Klimkowski, V. J. J. Phys. Chem. 1993, 97, 4528-4534. (17) Mattison, K. W.; Wang, Y.; Grymonpre´, K.; Dubin, P. L. Macromol. Symp. 1999, 140, 53-76. (18) Dubin, P. L.; Murrell, J. M. Macromolecules 1988, 21, 22912293. (19) Park, J. M.; Muhoberac, B. B.; Dubin, P. L.; Xia, J. Macromolecules 1992, 25, 290-295. (20) Mattison, K. W.; Brittain, I. J.; Dubin, P. L. Biotechnol. Prog. 1995, 11 (6), 632-637. (21) Mattison, K. W.; Dubin, P. L.; Brittain, I. J. J. Phys. Chem. B 1998, 102, 3830-3836. (22) Xia, J.; Dubin, P. L.; Dautzenberg H. Langmuir 1993, 9, 20152019. (23) Dubin, P. L.; Davis, D. D. Macromolcules 1984, 17, 1294-1296. (24) Kaibara, K.; Okazaki, T.; Bohidar, H. B.; Dubin, P. L. Biomacromolecules 2000, 1, 100-107. (25) Weinbreck, F.; de Vries, R.; Schrooyen, P.; de Kruif, C. G. Biomacromolecules 2003, 4, 293-303. (26) Girard, M.; Turgeon, S. L.; Gauthier, S. F. J. Agric. Food Chem. 2003, 51, 6043-6049. (27) Weinbreck, F.; Nieuwenhuijse, H.; Robijn, G. W.; de Kruif, C. G. Langmuir 2003, 19, 9404-9410. (28) Girard, M.; Turgeon, L.; Gauthier, S. F. Food Hydrocolloids 2002, 16, 585-591.

Langmuir, Vol. 21, No. 1, 2005 387

polymeric) complexes before phase separation occurs. It may be expected that in the complexation of biopolymers, more than two pH values corresponding to structural transitions will be identified. From this perspective and in the frame of a global research topic dealing with the study of β-lactoglobulin/Acacia gum/water (BLG/AG) complex coacervation,29-31 the present work aimed to determine the different scales of the structural transitions occurring in BLG/AG systems. This determination will provide also more insights on the significance of the pHφ, which is far from clear. Slow acidification of BLG/AG dispersions was obtained as a function of time by the hydrolysis of glucono-δ-lactone. This slow acidification coupled with different experimental methods allows the in situ determination of structural changes in mixed protein-polysaccharide dispersions.25,32 Dynamic light scattering, electrophoretic mobility, turbidimetry, circular dichroism, and phase contrast microscopy were used to characterize the pH-induced complexation and coacervation with decreasing pH. Materials and Methods Materials. β-Lactoglobulin powder (lot JE002-8-922) was provided by Davisco Foods International, Inc. (Lesueur, MN). The powder composition was (g per 100 g) 89.8% protein (N × 6.38), 8.8% moisture, and 1.4% ash. The mineral composition was (g per 100 g) 0.013 Mg2+, 0.079 Ca2+, 0.576 Na+, 0.097 K+, and 0.050 Cl-. Powdered Acacia gum (lot no. 97 J 716) was a gift from CNI Company (Rouen, France). The powder composition was (g per 100 g) 90% polysaccharide, 6.6% moisture, 0.3% nitrogen, and 3.1% ash. The mineral composition was (g per 100 g) 0.2 Mg2+, 0.61 Ca2+, 0.032 Na+, and 0.9 K+. Glucono-δ-lactone, GDL (lot B32827, Calbochiem, Darmstadt, Germany), is an ester of gluconic acid obtained by oxidative fermentation of D-glucose (Acetobacter suboxidans). Sodium hydroxide and hydrochloric acid were of analytical grade (Fischer Scientific SA, Elancourt, France). Preparation of β-Lactoglobulin and Acacia Gum Stock Dispersions. β-Lactoglobulin (BLG) and Acacia gum (AG) aqueous stock dispersions at 0.1 wt % final total biopolymer concentration were prepared first by dispersion of biopolymer powders in ultrapure deionized water (18 mΩ resistivity) (Purite, Fischer Scientific, England) under gentle stirring at 20 ( 1 °C for at least 2 h. The pH of the BLG dispersion was adjusted at 4.80 (pH corresponding to the lowest BLG solubility) with 0.1 and 1 N HCl. One hundred fifty microliters of 0.1 N NaOH was added to AG dispersion in order to reach a pH value of ∼7.00 after 18 h of equilibration time. The dispersions were stored at 4 ( 1 °C during 18 h to allow good hydration of biopolymers. The stock dispersions were centrifuged during 40 min at 10000g to remove insoluble BLG aggregates, insoluble matter, and air bubbles. The concentration of BLG dispersion was checked by spectrophotometry (Ultrospec 4000 UV/Visible, Pharmacia Biotech, England) at 278 nm, using the specific extinction coefficient of 9.6 dL‚cm-1‚g-1.33 Optical density at 278 nm was corrected for turbidity. The pH of the BLG dispersion was then adjusted at 7.20 using 0.1 and 1 N NaOH. Preparation of BLG/AG Mixed Dispersions. BLG/AG mixed dispersions were prepared by adding AG dispersion to BLG dispersion at a protein:polysaccharide weight ratio of 2:1. (29) Schmitt, C.; Sanchez, C.; Thomas, F.; Hardy, J. Food Hydrocolloids 1999, 13, 483-496. (30) (a) Schmitt, C.; Sanchez, C.; Despond, S.; Renard, D.; Thomas, F.; Hardy, J. Food Hydrocolloids 2000, 14, 403-413. (b) Schmitt, C.; Sanchez, C.; Lamprecht, A.; Renard, D.; Lehr, C. M.; de Kruif, C. G.; Hardy, J. Colloids Surf., B 2001, 20 (3), 267-280. (c) Schmitt, C. The`se de Doctorat, Institut National Polytechnique de Lorraine, France, 2000. (31) Sanchez, C.; Despond, S.; Schmitt, C.; Hardy, J. In Food Colloids: Fundamentals of formulation; Dickinson, E., Miller, R., Eds.; Royal Society of Chemistry: Cambridge, 2001; p 332. (32) Tuinier, R.; Rolin, C.; de Kruif, C. G. Biomacromolecules 2002, 3, 632-638. (33) Townend, R.; Winterbottom, R. J.; Timasheff, S. N. J. Am. Chem. Soc. 1960, 82, 3161-3168.

388

Langmuir, Vol. 21, No. 1, 2005

Mekhloufi et al.

The blend was gently stirred for 30 min. The pH of the blend was at 7.20 at the end of the stirring period. Mixed dispersions were filtered through a 0.22-µm syringe filter (Millipore, Bedford, MA). A 0.11% (wt/wt) portion of GDL was then added to the blend and mixed for 5 min. In aqueous medium, GDL is hydrolyzed until an equilibrium is reached between gluconic acid and GDL. Gluconic acid releases protons and gluconate in the medium. This proton release leads to a slow medium acidification. The structure of BLG/GA mixed dispersions as a function of pH was then investigated using dynamic light scattering, turbidimetry, electrophoretic mobility, circular dichroism, and phase contrast microscopy. The pH values of dispersions were recorded as a function of time for each experiment. Dynamic Light Scattering. A Malvern ZetaSizer III (Malvern Instrument, England) was used to follow the evolution of light scattering during acidification. The apparatus is equipped with a 5 mW He-Ne laser (633 nm), a measurement cell, a photomultiplier, and a correlator. Samples were placed in vertical cylindrical cuvettes (9 mm diameter). Scattered light intensity (counts‚s-1) at 90° scattering angle (I90) was recorded every 45 s during 75 min, at room temperature (20 ( 0.1 °C). The I90 value was subtracted by the intensity at t ) 0 min. Two independent experiments were performed. Measurement errors were less than 5% on characteristic pH values and 10.3% on I90 values at these pH. Size and Electrophoretic Mobility Measurements. Size measurements were carried out with a dynamic light scattering type Zetasizer3000 HAS (Malvern instruments, U.K.) apparatus, equipped with a 10 mW He-Ne laser (632.8 nm), photodiode detector, and laser attenuator. Vertical cuvettes with a path length of 10 mm were used as scattering cell. Measurements were performed at a scattering angle of 90°, at room temperature (23 ( 2 °C). Intensity autocorrelation functions were recorded every 30 s during 30 min and then analyzed by a CONTIN algorithm (integrated in the Malvern Zetasizer software) in order to determine the diffusion coefficients (or hydrodynamic radii) distribution. The diffusion coefficient D is related to the hydrodynamic radius (Rh) of particles through the StokesEinstein relationship

D ) kBT/6πηRh

(1)

where η is the solvent viscosity and kBT is the thermal energy. Results from the CONTIN algorithm allows the determination of both Rh distribution and respective amplitude of each population. Six independent experiments were performed for BLG/AG mixed dispersions. Three of them were performed without laser attenuation to check on multiple scattering effects. It was observed that Rh had almost the same evolution during acidification with and without laser attenuation. However, Rh values determined at the end of the experiments need to be considered with caution because of a significant turbidity in the dispersions. Three independent measurements were performed on control BLG and AG dispersions at 0.066 and 0.033 wt % total concentrations, respectively. Measurement errors were less than 5% on characteristic pH values and 18% on Rh values at these pH. Electrophoretic mobility measurements (µE) were performed on the same apparatus by the means of laser Doppler electrophoresis. The sample was put in a standard capillary electrophoresis cell equipped with platinum electrodes. The apparatus was previously calibrated with latex standard (-50 ( 5 mV). Measurements were recorded every 30 s during 30 min. Measurements on control AG dispersions (AG) were performed at 1 wt % to ensure a good signal-to-noise ratio. Measurements on control BLG dispersions (BLG) were performed at 0.066 wt % (BLG concentration in the blend). Three independent experiments had been performed for BLG/AG, AG and BLG dispersions. Measurement errors were less than 10% on characteristic pH values and 25% on electrophoretic mobility values at these pH. For the purposes of clarity, electrophoretic mobility units (µm‚cm‚s-1‚V-1) will be referred to as emu in the following. Turbidity Measurements and Phase Contrast Microscopy. An Ultrospec 4000UV/Visible spectrophotometer (Pharmacia Biotech, England) was used to follow the turbidity during acidification at a wavelength of 633 nm. Samples were placed in

a 10 mm path length plastic cuvette. Turbidity was recorded every 30 s during 60 min (at 20 ( 0.1 °C) and then were calculated as follows

τ ) (-1/L) ln(I0/It)

(2)

with L the optical path length (cm), It the transmitted light intensity, and I0 the incident light intensity. The turbidity τ at t ) 0 min was subtracted from the turbidity as a function of pH. Two independent experiments were performed. Measurement errors were less than 5% on characteristic pH values and 16% on turbidity values at these pH. A phase contrast microscope (Leica DMRB, Wetzlar, Germany) was used to monitor structural changes in BLG/AG dispersions during acidification. It was connected to a CCD video camera (Kappa optoelectronics GmbH, Germany) controlled by an image processor (Kappa ImageBase 2.5). A 30-µL sample was taken from BLG/AG mixed dispersion at different pH values (corresponding to different times in the acidification kinetic) and put between glass slides at room temperature (22 ( 1 °C). Pictures were taken at ×40 magnification. The contrast and the brightness of the picture recorded at pH 4.2 (Figure 4b in the section Results) were processed by the Scion Image freeware (version Beta 4.0.2 by Scion Corporation) in order to better visualize light fluctuations. Two independent measurements were carried out. Circular Dichroism Measurements. Far-UV circular dichroism was used at 20 ( 0.2 °C and 208 nm wavelength using a Jobin-Yvon CD6 Spex dichrograph purged by nitrogen flow. On the basis of preliminary dichroic spectra (15 scans and 164 s for each scan), the 208 nm wavelength was the most sensitive wavelength in the present study to detect protein conformational changes. The absorbance variation (∆A) as a function of pH was recorded on BLG, AG, and BLG/AG dispersions every 2 s during 42 min. Samples were placed in quartz cell with a path length of 0.01 cm. The ∆A of BLG alone (which was not involved in interaction with AG) and BLG in the mixed dispersion (which interacted with AG) were considered and compared. ∆A of BLG alone was obtained by subtracting the GDL and solvent signals from the BLG signal. ∆A of BLG in the mixed dispersion was obtained subtracting the GA, GDL, and solvent signals from the BLG/AG dispersions signal. The ellipticity data (in deg) were then converted into mean residue ellipticity (in deg‚cm2‚dmol-1) using a Mw for BLG of 36 800 g‚mol-1 (dimeric form). The relative molar ellipticity of BLG per residue at 208 nm ([θ]MRW/[θ]MRW(t)0)) was calculated. Three independent experiments were performed on each sample. Measurement errors were less than 5% on characteristic pH values and 10% on relative molar ellipticity values at these pH. Determination of Structural Transition pH. The pH values corresponding to the different structural transitions were determined graphically for each method as the intersection point of two tangents to the curve. Average and standard deviations were calculated for each determined pH.

Results Identification of Characteristic pH Transitions in a Complex Coacervation Process. pH-induced structural transitions occurring in the BLG/AG mixed dispersion were determined by using complementary methods. Light scattering at 90° scattering angle (I90) is sensitive to refractive index differences at the length scale of macromolecules and macromolecular assemblies of limited sizes. Turbidity (τ) is sensitive to large scale refractive index inhomogeneities.34 Size measurements give information on the evolution of the average hydrodynamic radius (Rh) of particles determined by the peak position of the Rh distribution curves. Five pH-delimited zones were determined using these different methods (Figures 1-3): (34) Urwin, J. R. In Molecular weight distributions by turbidimetric titration in light scattering from polymer solutions; Huglin, M. B., Eds.; Academic Press: New York, 1972; Chapter 18.

pH-Induced Structural Changes

Langmuir, Vol. 21, No. 1, 2005 389

Figure 1. pH-induced evolution of (b) the scattered light intensity at 90° angle (I90) and (O) the turbidity (τ) as a function of pH for the BLG/AG mixed dispersions at 0.1 wt % total biopolymer concentration and BLG:AG weight ratio of 2:1. Inset: (b) (dI90/dt) (counts‚s-2) and (O) (dτ/dt) (×10-3 cm-1‚s-1) evolution as a function of pH for the same system. Each point is the average of two independent experiments. Arrows indicate the pH of structural transitions.

Figure 3. Size distributions in intensity of hydrodynamic radii for the BLG/AG mixed dispersions at 0.1 wt % total biopolymer concentration and a BLG:AG weight ratio of 2:1. Each point is the average of three independent experiments. Arrow indicates the pH evolution.

Figure 2. (a) Hydrodynamic radii evolution as a function of pH for the BLG/AG mixed dispersions at 0.1 wt % total biopolymer concentration and BLG:AG weight ratio of 2:1. (b) Enlargement of (a) at the first pH values. Each point is the average of six independent experiments. Arrows indicate the pH of structural transitions.

pH 5.7 to pH ∼ 4.90: the I90 and τ values were low and constant (Figure 1). At this stage, both biopolymers coexisted in the mixed dispersion but apparently did not interact. Rh was almost constant at ∼40 nm (Figure 2a). This value corresponded to the averaged Rh of AG, the average BLG radius being too small (∼2 nm) to be detected (results not shown). According to Figure 3, hydrodynamic radii of the BLG/GA mixed dispersions displayed bimodal distribution at pH 5.61 and 4.92. The first peak was

centered near 35 nm and the second one was beyond 1000 nm. Mixed dispersions were highly polydispersed at the beginning of acidification. pH ∼ 4.90 to pH 4.50: at pH 4.93 ( 0.02 for I90 and pH 4.90 ( 0.02 for τ, recorded parameters significantly departed from the baseline. Rh gradually increased from pH 4.88 ( 0.11 (Figure 2). These events suggested the formation of local structural inhomogeneities in the BLG/ AG mixed dispersion, as a result of complexation between BLG and AG. Structural entities continued to grow in the mixed dispersion through acidification, as revealed by an increasing I90 (with a constant rate in the pH range from 4.79 ( 0.02 to 4.52 ( 0.01, see inset in Figure 1) and a slight increase of the turbidity. The polydispersity of the mixed dispersion considerably diminished from pH 4.92 and 4.68 (Figure 3), as can be seen by the difference of size distribution width and the disapperance of the larger

390

Langmuir, Vol. 21, No. 1, 2005

Mekhloufi et al.

Figure 4. Phase contrast micrographs at different pH values for the BLG/AG mixed dispersions at 0.1 wt % total biopolymer concentration and BLG:AG weight ratio of 2:1. (a) pH 4.90, (b) pH 4.20, (c) pH 4.00, (d) pH 3.90, (e) pH 3.80, and (f) pH 3.75. (g) A focused region of the micrograph (f) illustrating coalescence (white arrows). Bar represents 10 µm. Contrast and brightness of micrograph b were treated by the Scion Image freeware (version Beta 4.0.2, Scion Corp.).

particles (Rh ∼ 1 µm). When the pH ranged from 4.68 to 4.63, Rh suddenly increased from ∼68 to ∼88 nm (Figure 2b), indicating a clustering mechanism. pH 4.50 to pH 4.20: in this pH range, τ and Rh continued to increase steadily, whereas I90 strongly increased with a maximum evolution rate at pH 4.35. Structural entities were then more numerous and larger, their Rh increased from ∼100 nm (at pH 4.50) to ∼155 nm (at pH 4.20), with a narrow size distribution (Figure 3). pH 4.20 to pH 4.04: The maximum of I90 was observed at pH 4.20 ( 0.01 indicating a maximum number of complexed structural entities. On the other hand, the increase of τ considerably accelerated from pH 4.23 ( 0.02, with a maximum evolution rate at pH 4.11 (inset in Figure 1). Rh underwent also a significant increase at pH 4.21 ( 0.04 (Figure 2b), resulting in the increase of particle polydispersity (note the difference in distribution width with pH 4.12 in Figure 3). Almost all particles displayed a Rh larger than 100 nm below pH 4.2. These results all revealed that pH 4.2 was an important structural transition in the BLG/AG system. pH 4.04 to pH 3.8: τ reached a maximum at pH 4.04 ( 0.01 and then gradually decreased. Below pH 4.04, the number of structural entities was reduced, either because of a pH-induced dissociation of the structural entities or because of a size increase of these entities to the detriment

of their number. The latter hypothesis is more likely because of the Rh increase (>250 nm) observed beyond pH 4.04. pH-Induced Microstructural Evolution of BLG/ AG Mixed Dispersions. Phase contrast microscopy is a useful method to investigate the structure of phaseseparating mixed dispersions. Figure 4 shows micrographs taken at different pH values during acidification of BLG/ AG mixed dispersions. No structure was detected before pH 4.20 (Figure 4a,b). At pH 4.20, very small light fluctuations were observed on the monitor screen indicating that phase separation started (Figure 4b). Maximum light fluctuations, as well as some first particles, were observed near pH 4.00, in the vicinity of the pH of the maximum of turbidity (Figure 4c). Nearly 1 µm coacervates (spherical droplets) were clearly distinguished at pH 3.90 (Figure 4d). It can then be inferred that coacervation occurred in the 4.0-3.9 pH range. Coacervates grew in size during acidification, and their number was reduced (Figure 4e,f). This feature could be due to coalescence of coacervates (Figure 4g) or Ostwald ripening and also probably to sedimentation of the largest particles. It is important to note, as suggested by one of the reviewer, that sample confinement between glass slides could affect kinetics of phase separation. In the present study, the confinement was limited by taking samples at different pH values directly from the beaker; i.e., structuring

pH-Induced Structural Changes

Figure 5. Relative molar ellipticity of BLG per residue at 208 nm ([θ]MRW/[θ]MRW(t)0)) as a function of pH for the BLG/AG mixed dispersions at 0.1 wt % total biopolymer concentration and BLG: AG weight ratio of 2:1. Inset: [θ]MRW/[θ]MRW(t)0) as a function of pH for the BLG control dispersion. Each point is the average of three independent experiments.

kinetics was not followed in situ, putting them between glass slides and taking micrographs as rapidly as possible. pH-Induced Conformational Changes of BLG in BLG/AG Dispersions. In addition to the previous methods, circular dichroism (CD) allowed the determination of the pH-induced conformational transitions of BLG during the complex coacervation process. Figure 5 shows the evolution of the relative molar ellipticity of BLG per residue at 208 nm ([θ]MRW/[θ]MRW(t)0)) as a function of pH for the BLG/AG mixed dispersion. Three main pH transitions were observed. The relative molar ellipticity at 208 nm was constant from pH 5.68 to pH 4.77 ( 0.01. BLG secondary structure did not change in this pH range. From pH 4.77 ( 0.01, the relative molar ellipticity of BLG decreased and pointed out mainly a loss of R-helix structure. The loss of BLG secondary structure continuously increased until pH 3.96 ( 0.02 was reached, with a more marked BLG conformational change at pH 4.35 ( 0.03. Below pH 3.96 ( 0.02, the relative molar ellipticity increased, indicating a gain in BLG secondary structure and suggesting a molecular reorganization of the protein. Under the same experimental conditions, the relative molar ellipticity at 208 nm for the BLG control dispersion showed a slight increase during acidification (inset in Figure 5), indicating a minimal gain of secondary structure. No changes in relative molar ellipticity were detected for the AG control dispersion (results not shown). Electrophoretic Mobility of BLG/AG Complexes and BLG and AG Macromolecules. As complex coacervation between proteins and polysaccharides is mainly controlled by electrostatic interactions, electrophoretic mobility (µE) measurements were used to extract information on total surface charges of BLG, AG, and complexed BLG/AG. Figure 6 represents the pH-induced changes of BLG, AG, and BLG/AG µE. The BLG µE changed from -1.05 emu at pH 5.6 to +0.34 emu at pH 4.27, depending on the charge balance between the amino and the carboxylic groups carried by the protein. The zero µE

Langmuir, Vol. 21, No. 1, 2005 391

Figure 6. Electrophoretic mobility as a function of pH: (b) BLG/AG mixed dispersions at 0.1 wt % total biopolymer concentration and BLG:AG weight ratio of 2:1; (O) BLG dispersions at 0.066 wt % protein concentration; (0) AG dispersions at 1 wt % gum concentration. Each point is the average of three independent experiments. Arrows indicate the pH of structural transitions. Solid lines are just guides for the eyes.

value of the BLG was reached at pH 4.85 ( 0.05, below the isoelectric point of the protein (∼5.1) and as expected closely corresponded to the lowest BLG solubility (see Materials and Methods). The AG µE was negative in the pH range considered in the present study (from 5.17 to 3.98), due to the presence of glucuronic acid residues in the gum (pKa < 3.6). The total charges of AG slightly decreased with decreasing pH (µE decreased from -1.95 emu at pH 4.90 to -1.34 emu at pH 3.93), as less acidic COOH groups became ionized. In the BLG/AG mixed dispersion, µE was also negative in the 5.60-3.96 pH range and displayed a similar whole evolution than that of Acacia gum alone. This highlighted the main contribution of AG to the µE evolution of the mixed dispersion. An increase of the number of negative charges was first observed between pH 5.60 (-1.5 emu) and pH 5.11 ( 0.08 (-1.7 emu). µE then slowly decreased until a pH value of 4.41 ( 0.04 was reached. From this pH, total negative charges in the mixed dispersion decreased faster since complexes went closer to electroneutrality. Discussion Acidification of BLG/AG mixed dispersions led to the formation of coacervate type microscopic droplets. Combining the signal evolution of all methods described above, different structural transitions from the molecular to the microscopic scale were identified during the phase transition. The general pH-induced complex coacervation process between BLG and AG was divided into different stages delimited by six pH ranges corresponding to particular phenomena and structural changes (see definitions Table 1). Above pH 4.90, BLG and AG coexisted in the mixed dispersion and apparently did not interact due to the repulsive Coulombic forces between the two negatively charged biopolymers. The average Rh in the mixed dispersion was about 40 nm, which corresponded to the averaged size of the Acacia gum measured in the control AG dispersion. Weinbreck et al.25 also remarked that the particle size in a whey protein/Acacia gum mixture,

392

Langmuir, Vol. 21, No. 1, 2005

Mekhloufi et al.

Table 1. Summary of the pH-Induced Transitions in the BLG/AG Mixed Dispersions at 0.1% Total Biopolymer Concentration and BLG:AG Weight Ratio of 2:1 Determined for Each Measured Parametera measured parameters µE I90 dI90/dt τ dτ/dt Rh [θ]MRW/[θ]MRW(t)0) av pH

pHsc

pH of structural transitionsb pHpct pHca pHpsi pHps

pHcoa

4.41 4.93 4.93 4.90 4.90 4.88 4.9

4.20 4.77

4.35 4.44 4.68

4.77 4.8

4.7

4.45 4.35 4.4

4.23

4.07 4.04 4.00

4.21 4.2

3.96 4.0

a Key: µ , electrophoretic mobility (emu, µm‚cm‚s-1‚V-1); I , E 90 scattered light intensity at 90° scattering angle (counts‚s-1); dI90/ dt, first derivative of I90 (counts‚s-2); τ, turbidity (cm-1); dτ/dt, first derivative of turbidity (cm-1‚s-1); Rh, hydrodynamic radius (nm); [θ]MRW/[θ]MRW(t)0), relative molar ellipticity per residue). b Key: pHsc, pH of formation of soluble complexes; pHpct, pH of protein conformational change; pHca, pH of complex aggregation; pHpsi, pH of phase separation initiation; pHps, pH of phase separation; pHcoa, pH of coacervate formation.

determined by DLS at 90° scattering angle, was similar to the size of Acacia gum alone determined in the blank dispersion (∼125 nm) in the pH range where no interactions between macromolecules occurred. However, the measured size of Acacia gum was three times higher than that used in our study. Our AG sample displayed a polydisperse size distribution (Rh from ∼5 to >400 nm). This high polydispersity, previously described in the literature,30c,35-37 can be explained in part by the complex molecular composition of Acacia gum. It contains three main molecular species, an arabinogalactan, an arabinogalactan-protein, and a glycoprotein. The arabinogalactan fraction represents 88 wt % of the whole gum and has a ∼10 nm hydrodynamic radius. The arabinogalactan-protein fraction represents 10 wt % of the gum, and its hydrodynamic radius is about 30 nm. The polydispersity (Mw/Mn) of our AG sample was 2.3 as determined experimentally.36 The difference between both studies could not be explained solely, in our opinion, by the strong biological variability of AG samples.36 As an example, the intrinsic viscosity [η] of 75 Acacia gum samples was shown to be between 0.14 and 0.60 dL‚g-1.36 More likely, differences could be explained on the basis of different AG sample preparation (e.g., powder dissolution time and dispersion equilibration time, dispersion filtration) that could induce different AG solubility and the presence of aggregated fractions of Acacia gum. The upper measured sizes in the present study would correspond to these aggregated fractions.30c The wide size distribution of the AG molecule consequently contributed to the high polydispersity of the BLG/AG mixed dispersion above pH 4.9. In addition, the control BLG dispersion displayed a bimodal size distribution with two peaks located at ∼7 and >130 nm. The second peak corresponded to residual BLG aggregates, which represented 20% of the scattering population at pH 5.45 (results not shown). The total net charge in the mixed dispersion was negative. An increase in the number of negative charges was observed in the mixed dispersion between pH 5.6 and pH 5.11. Since, in the same pH range, BLG µE became less (35) Picton, L.; Bataille, I.; Muller, G. Carbohydr. Polym. 2000, 42, 23-31. (36) Sanchez, C.; Renard, D.; Robert, P.; Schmitt, C.; Lefebvre, J. Food Hydrocolloids 2002, 16, 257-267. (37) Idris, O. H. M.; Williams, P. A.; Phillips, G. O. Food Hydrocolloids 1998, 12, 379-388.

negative and tended to electroneutrality, the gain of negative charges in the mixed dispersion was due entirely to chemical or structural changes affecting AG. The most likely explanation concerns the dissociation of AG aggregates leading to the unmasking of buried negatively charged groups. The disappearance of bimodal size distribution of mixed dispersions below pH 4.92, corresponding to the loss of the peak centered at about 1000 nm, supports this assumption. Because of the fast acidification of AG dispersions by GDL that precluded µE measurements above pH 5.2, it was not possible to know from our measurements whether the dissociation of AG aggregates was only induced by pH, or by a combined action of pH and BLG. In a previous study, Burgess11,38,39 observed that the negative µE of Acacia gum increased (became more negative) from pH 2.2 to pH 4.5 and then remained constant until pH 10. Hence, it seems that mixing BLG with AG leads to a dissociation of AG. A likely hypothesis is that the added counterions brought by BLG dispersion promoted AG dissociation. BLG/AG soluble complexes induced by electrostatic interactions were formed at approximately pH 4.90, which was the critical pH, pHc, according to the terminology adopted from the literature. However, as evidenced in the present study, other critical pH values exist during complexation/coacervation of biopolymers. To avoid any confusion, we named pH 4.90 pHsc, the pH of soluble complex formation. According to electrophoretic mobility measurements, AG was strongly negatively charged at pH 4.9 whereas BLG displayed a very small excess of negative charges (zero electrophoretic mobility at pH 4.85). Complexation in these conditions could be explained by the fact that BLG carries on its surface positive charge density areas, called “patches”, that would enable local interactions between BLG and AG above or close to the isoelectric point of the protein.19 A “patch” would be composed of five positively charged amino acid residues (Lys135,138,141, His146, and Arg148), making part of the outer main R-helical structure of BLG.40 An interaction between this patch and the negatively charged carboxylic groups of Acacia gum could explain the loss of R-helix observed for the protein at pH 4.77. This latter was then defined as the pH value corresponding to protein conformational transition and was named pHpct. It was previously shown that a partial conformational change of the BLG helical region could be observed upon interaction with Acacia gum.41 The small difference between pHsc and pHpct could be explained, either by conformational changes occurring slower than the acidification rate or by kinetic slowing down induced in the very thin CD optical cell (0.01 mm). It is important to note that a consequence of biopolymer complexation was the progressive homogenization of particle sizes in the mixed dispersion. The size of the formed complexes grew until pH 4.68, from which Rh suddenly increased and polydispersity decreased. Particles evolved to larger size, and no particles smaller than 30 nm were detected. pH 4.68 was then the pH where soluble complexes interacted to form aggregated complexes and was named pHca, the pH of complex aggregation. One possible reason is that the density of (38) Burgess, D. J.; Kwok, K. K.; Megremis, P. T. J. Pharm. Pharmacol. 1991, 43, 232-236. (39) Burgess, D. J.; Carless, J. E. J. Colloid Interface Sci. 1984, 98, 1-8. (40) Molinari, H.; Ragona, L.; Varani, L.; Musco, G.; Consonni, R.; Zetta, L.; Monaco, H. L. FEBS Lett. 1996, 381, 237-243. (41) Schmitt, C.; Sanchez, C.; Despond, S.; Renard, D.; Robert, P.; Hardy, J. In Food Colloids: Fundamentals of formulation; Dickinson, E., Miller, R., Eds.; Royal Society of Chemistry: Cambridge, 2001; p 323.

pH-Induced Structural Changes

Langmuir, Vol. 21, No. 1, 2005 393

positive charges onto BLG becomes sufficient to overcome repulsion between the complexed proteins and part of the AG molecules involved in other complexes. Aggregation of BLG/AG complexes was very important in the pH region between 4.68 and 4.2. This pH range appears essential since mixed dispersion acceded a higher structural level. Two major structural transitions appeared at, pH 4.4 and pH 4.2, respectively. Around pH 4.4, the rate of I90 evolution decreased, the turbidity began to significantly increase, and a more marked BLG conformational change was observed. These structural events suggest that large scale aggregation of complexes with molecular reorganization of protein occurred from this pH. Since the number of BLG positive charges leveled off and the number of AG negative charges further decreased, stronger charge neutralization of aggregated complexes promoted further clustering. BLG-BLG interactions at pH 4.4 ( 0.05, as determined by a slight variation of I90 of control BLG dispersion (result not shown), could also contribute to the observed structuring acceleration. Carlsson et al.42 studied by Monte Carlo simulations the complexation in model systems composed of protein models (representing lysozyme in aqueous environment) and negatively charged polyelectrolyte. The authors showed that nonelectrostatic protein-protein attraction promoted protein-polyelectrolyte association and facilitated proteinpolyelectrolyte cluster formation. This effect could be explained by considering both entropic (mainly due to the polydispersity in size and in molecular mass) and enthalpic (mainly due to the surface properties of aggregates) factors.30a,43 From pH 4.2, I90 attained its maximum value, τ displayed a very important acceleration, Rh strongly increased, particle polydispersity increased, and small light fluctuations (like a dark vibrating cloud) were observed by phase contrast microscopy. All these observations suggested that phase separation was occurring. Mixed dispersions then became unstable in the 4.4-4.2 pH range that was the transition pH region between BLG/ AG complexation and coacervation. When defining pH-induced structural transitions, and comparing our data to those reported in the literature, one question arises: Which pH value corresponds to the pHφ (see Introduction)? In fact, this depends on the pHφ definition that is not always the same depending on authors and publication year. If the pHφ corresponds to the first important increase in turbidity or to the first major structural events leading to phase separation, then pHφ is 4.4. If the pHφ corresponds to phase separation, pH 4.2 is the pHφ. To possibly clarify the situation, we infer that aggregation of complexes occurred during acidification until sufficient decrease of total negative charge (see below) promoted rapid clustering of complexes aggregates at the sub-microscopic scale (pH 4.4), inducing the destabilization of mixed dispersion. Phase separation then occurred (pH 4.2). In this frame, pH 4.4 was defined as the pH of phase separation initiation, i.e., corresponding to the first structural events leading to phase separation, and was named pHpsi. The pH of phase separation was 4.2 and was named pHps. In our opinion, such a distinction is coherent with experimental results that show structural transitions at two structural levels. We also propose that in the future, pHps for biopolymer mixtures will be determined not only by an often subjective “significant” increase in turbidity but also on the basis of the following characteristics: I90 maximum (as suggested by Kaibara

et al.24), considerable acceleration of the turbidity (through the dτ/dt), marked decrease in aggregated complexes electrophoretic mobility, and appearance of microscopic concentration fluctuations. Phase separation is often described as the consequence of an insolubilization of aggregated complexes. It was previously found that phase separation occurred when the net charge of interpolymer complexes was close to 0.18 We have to remember that, besides entropic contributions, stability of mixed biopolymer dispersions is much determined by the balance between biopolymer-biopolymer and biopolymer-solvent interactions. According to the Veis-Aranyi theory,15 attractive interactions between biopolymers and low biopolymer-solvent interactions are favorable to coacervation. When neutral aggregated complexes are formed, as a consequence of strong attractive interactions, aggregated complexes-solvent interactions become low and phase separation occurs. Nevertheless, Xia et al.16 found that the requirement of neutrality at phase separation was not a general rule and was only observed for complexes formed by protein-polycationic polymers. For protein-polyanionic polymers, phase separation occurred for complexes with a negative total charge. No explanation was provided. The same result was observed in the present study. In fact, solubility of complex aggregates decreased, as evidenced by the continuous decrease of negative electrophoretic mobility, but electroneutrality was not reached even at pHPS. Interesting features from electrophoretic measurements were that (i) negative charge of AG was constant from pH 4.9 to 4.5 whereas BLG was increasingly positively charged and (ii) positive charge of BLG was almost constant from pH 4.5 to 4.2 whereas AG was less negatively charged, especially from pH 4.4. Therefore, the neutralization of AG negative charges by positively charged BLG could explain BLG/ AG complexation (many proteins interact with one AG molecule) and first complex aggregation events (the increase of positive charges favors interactions of complexed BLG with other AG molecules, involved or not in complexes). On the other hand, higher order aggregation of complexes and phase separation were controlled in part by charge neutralization but mainly by the decrease of AG negative charges, due to the closer proximity of pH to the pKa of polysaccharide glucuronic acids. If we suppose that the surface of aggregates contains a great proportion of the surface active and hydrophilic AG36 (compare the µE evolution in mixed dispersion to that in AG, see Figure 6), in this case the large scale aggregation of complexes could be due to a decrease of electrostatic/steric repulsion between the less charged/hydrated AG molecules present at the surface of the different aggregates. We have shown that this mechanism plays also an important role in the stability of BLG/AG coacervates.44 Once aggregates attain a critical size and (AG-induced) solubility, phase separation can occur. Leisner and Imae have recently shown that the missing electrostatic repulsion among surface of interpolyelectrolyte complexes and their high molecular weight favor the colloidal aggregation that precedes coacervation.45 Destabilization of mixed dispersion then seems to be controlled by a critical size/charge ratio parameter. This could explain that large scale aggregation of complexes and coacervation can occur without the requirement of aggregate neutrality. Concentration fluctuations were maximum in mixed dispersion near pH 4, where the turbidity was maximum,

(42) Carlsson, F.; Malmsten, M.; Linse, P. J. Am. Chem. Soc. 2003, 125, 3140-3149. (43) Overbeek, J. T. J.; Voorn, M. J. J. Cell. Comput. Physiol. 1957, 49, 7-26.

(44) Sanchez, C.; Mekhloufi, G.; Schmitt, C.; Renard, D.; Robert, P.; Lehr, C. M.; Lamprecht, A.; Hardy, J. Langmuir 2002, 18 (26), 1032310333. (45) Leisner, D.; Imae, T. J. Phys. Chem. B 2003, 107, 8078-8087.

394

Langmuir, Vol. 21, No. 1, 2005

and the first visible coacervates appeared. Coacervation of the whole system occurred in the 4.0-3.9 pH range since only coacervates were observed at pH 3.9. pH 4 was defined as the pH of coacervate appearance in the mixed dispersion at a microscopic level and was named pHcoa. From this pH, a gain in secondary structure of BLG was deduced from circular dichroism measurements. Knowing that turbidity of dispersions is a limiting parameter for CD analysis, it is important to understand whether the increase of [θ]MRW/[θ]MRW(t)0) was promoted by the BLG reorganization or simply caused by scattering effects. In the present study, [θ]MRW/[θ]MRW(t)0) decreased when the turbidity increased in the pH range from 4.77 to 4. In addition, the sampling cell had a thin path length, which contributed to strongly minimize light scattering. As well, [θ]MRW/[θ]MRW(t)0) increased with decreasing pH below pH 4 while turbidity in the sample decreased. The CD signal evolution could thus be interpreted unambiguously as a protein conformational change in the mixed dispersion independently of the turbidity level. To better understand BLG structural reorganization, it is useful to note that very recent experiments by static light scattering45 and small-angle X-ray scattering46 have revealed that complex aggregates display a condensation process during coacervation. One consequence of such a condensation process is that protein molecules could be structurally arranged in a compact manner in coacervates,46 i.e., locally more concentrated. An increase in the local concentration of BLG could induce conformational changes, especially an increase of protein aggregation, a feature observed on BLG solutions of increasing concentration by FTIR spectroscopy.47 Since protein aggregation involves in a great number of cases a loss of R-helix and an increase of intermolecular β-sheet structures, it is possible that the increase in the relative molar ellipticity per residue observed near pHcoa was strongly influenced by changes in β-sheet structures. This could be verified using FTIR spectroscopy that is more sensitive than circular dichroism to this kind of secondary structure. Conclusion The phase separation phenomenon through complex coacervation between BLG and AG was observed combining slow in situ acidification of the mixed dispersion and light scattering, spectroscopic, and microscopic methods. Different critical structural transitions and their corresponding pH values were highlighted. The interaction between BLG and AG was initiated by electrostatic interactions at pHsc, the pH of soluble complexes formation. At pHca, complex aggregation occurred, which rapidly clustered at the sub-microscopic scale at pHpsi, inducing the destabilization of the mixed dispersion. Phase separation then occurred at pHps. The pH range between pHpsi and pHps was the transition region between BLG/AG complexation and coacervation. The mixed dispersion destabilization seemed to be controlled by a critical size/ charge ratio parameter, which could explain the occurrence of phase separation while complex aggregates were still negatively charged. It was also deduced that complexation (46) Weinbreck, F.; Tromp, R. H.; de Kruif, C. G. Biomacromolecules 2004, 5, 1437-1445. (47) Czarnik-Matusewicz, B.; Murayama, K.; Wu, Y.; Uzaki, Y. J. Phys. Chem. B 2000, 104, 7803-7811.

Mekhloufi et al.

between BLG and AG was mainly controlled by neutralization of negative charges of AG by positive charges of BLG, whereas coacervation was mainly controlled by charge neutralization of complex aggregates through the lowering of AG negative charges. At a microscopic level, important concentration fluctuations and first coacervates appeared in the mixed dispersion at pHcoa, near the pH of the maximum of turbidity. Phase contrast microscopy also allowed a better understanding of structuring and aging phenomena in mixed dispersions. From the onset of interactions between biopolymers, the pH decrease led to a gradual homogenization of particle size in the mixed dispersion (as the polydispersity of the mixed dispersion decreased). Complex coacervation between BLG and AG induced protein conformational transitions. A loss of R-helix structure of the protein was observed at pHpct, in parallel or after the initiation of interactions between BLG and AG, and a gain in protein secondary structure occurred near pHcoa, probably induced by protein concentration in the coacervate. The gain of negative charges above pHSC and the disappearance of bimodal size distribution in the initial mixed dispersion (loss of the Rh peak centered around 1000 nm) supported the hypothesis of a BLGinduced dissociation of AG aggregates. Although the present study tried to better define pHinduced structural transitions at different scale in a complex biopolymer systems, the use of complementary methods would be useful to better understand the structural transitions, both kinetically and at the molecular level. A great challenge concerns in particular the crucial transition region between complexation and phase separation. More detailed information on the structure of the different entities formed and energetic of assembly are also needed, using for instance high-resolution cryoTEM microscopy and modern calorimetric techniques. However, more fundamental knowledge of complexation and coacervation between biopolymers will not be possible without a better knowledge of the molecular structure of the different AG molecular fractions and the use of low polydispersity samples. Purification of these molecular fractions has been very recently performed. Chemical and physical characterization of fractions is in progress. The next step will be to apply the multimethodological approach presented in this paper to mixed dispersions of BLG and the different molecular fractions of AG. Nomenclature AG: Acacia gum BLG: β-lactoglobulin GDL: glucono-δ-lactone pHsc: pH of formation of soluble complexes pHpct: pH of protein conformational change pHca: pH of complex aggregation pHpsi: pH of phase separation initiation pHps: pH of phase separation pHcoa: pH of coacervate formation I90: scattered light intensity at 90° scattering angle (count‚s-1) dI90/dt: first derivative of I90 (count‚s-2) µE: electrophoretic mobility (emu: µm‚cm‚s-1‚V-1) [θ]MRW/[θ]MRW(t)0): relative molar ellipticity per residue Rh: hydrodynamic radius (nm) τ: turbidity (cm-1) dτ/dt: first derivative of turbidity (cm-1‚s-1) LA0486786