Prevention of peptide fouling on ion-exchange membranes during electrodialysis in overlimiting conditions

Journal of Membrane Science 543 (2017) 212–221

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Prevention of peptide fouling on ion-exchange membranes during electrodialysis in overlimiting conditions

MARK

Mathieu Persicoa,b, Sergey Mikhaylina,b, Alain Doyena, Loubna Firdaousc, Victor Nikonenkod, ⁎ Natalia Pismenskayad, Laurent Bazineta,b, a

Institute of Nutrition and Functional Foods (INAF) and Department of Food Sciences, Université Laval, Québec, QC, Canada G1V 0A6 Laboratoire de Transformation Alimentaire et Procédés ElectroMembranaires (LTAPEM, Laboratory of Food Processing and ElectroMembrane Processes), Québec, QC, Canada GIV 0A6 c Agro-food and Biotechnology research Institute “Charles Viollette”, ProBioGEM Laboratory, UPRES-EA 1026, Polytech'Lille/IUT A, Université de Lille Nord de France, Avenue Paul Langevin, 59655 Villeneuve d'Ascq Cedex, France d Physical Chemistry Department, Kuban State University, Krasnodar, Russia b

A R T I C L E I N F O

A B S T R A C T

Keywords: Demineralization Peptide fouling Electrostatic interactions Water splitting Vortices

Peptide fouling occurring on anion- (AEMs) and cation-exchange membranes (CEMs) is one of the most serious issues of conventional electrodialysis (ED) process for hydrolysate demineralization. Nonetheless, recent studies discussed the advantages of non-conventional ED phenomena such as water splitting and electroconvection on decreasing scaling and fouling. Thereby, peptide fouling was characterized using four different ED regimes: no current applied, underlimiting (conventional), limiting (water splitting) and overlimiting (electroconvection and water splitting) conditions. Results demonstrated that fouling-related interactions were mainly electrostatic with AEMs whereas they were both electrostatic and hydrophobic with CEMs. After 60 min, the demineralization rate was six times higher in overlimiting than underlimiting conditions. In addition, peptide fouling was 62% and 36% lower in overlimiting condition for AEMs and CEMs, respectively. It was hypothesized that (1) water splitting would have repealed the peptide charges through its "barrier effect" and (2) electroconvective vortices generated at the membranes interfaces would have washed-out their surfaces and hampered the attachment of peptides. Interestingly, ED under overlimiting conditions is a promising way to avoid peptide fouling. Consequently, membranes lifetime would be longer and new ED applications would be possible.

1. Introduction Conventional electrodialysis (ED) is a technology reported for various applications in chemical, pharmaceutical and food industries [1–3]. Nonetheless, fouling and scaling occurring at the interface of ionexchange membranes (IEMs) are the main limitations of ED since they diminish the process efficiency. Recent studies discussed the fouling mechanisms of complex peptides mixture and showed that it can occur on the IEMs mainly through the establishment of electrostatic interactions without application of current [4,5]. Moreover, fouling quantity is ruled by the pH of the environment since it determines the net charge of peptides. However, the literature provides only information concerning peptide fouling in static conditions and does not deal with the application of current in real hydrodynamic conditions. Furthermore, no information is available concerning the influence of non-conventional currents (limiting and overlimiting regimes) compared to the conventional one (underlimiting) on the peptide fouling. Indeed, since recent

⁎

studies showed the advantages of working in non-conventional modes on organic matter and protein fouling of IEM, concentration polarization decrease and mass transfer improvement, further investigations should be carried-out for peptide fouling [6,7]. However, in limiting and overlimiting conditions, some phenomena appear at the membrane interfaces hampering mass transfer and pH evolution. In limiting conditions, water molecules are dissociated into protons and hydroxyl ions at the IEM interface causing pH variations [8]. Consequently, the pH-dependent charges of peptides may affect fouling [4,5]. Moreover, Bukhovets et al. [9] demonstrated that pH modifications would wash out phenylalanine residues previously stuck into an AEM matrix and therefore diminish fouling. In overlimiting conditions (intensive applications of tension), there is a development of electroosmosis of the second kind, which creates an excessive pressure within the space charged region [7]. This phenomenon leads to the formation of electroconvective vortices which accelerate the migration of ionic species through the IEM by mixing the depleted solution

Corresponding author at: Institute of Nutrition and Functional Foods (INAF) and Department of Food Sciences, Université Laval, Québec, QC, Canada. E-mail address: [email protected] (L. Bazinet).

http://dx.doi.org/10.1016/j.memsci.2017.08.039 Received 26 May 2017; Received in revised form 14 August 2017; Accepted 17 August 2017 Available online 23 August 2017 0376-7388/ © 2017 Elsevier B.V. All rights reserved.

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

[10,11]. Besides, Mikhaylin et al. [12]. showed that such vortices would limit the scaling formation on CEMs. Furthermore, the overlimiting mode requires a lower surface of IEMs to treat the same quantity of solution comparing to the underlimiting one. Since the cost of IEMs is substantial in the total budget, the overlimiting mode may be financially more attractive despite the growing costs of the related electric consumption [13]. In this context, this work aimed to investigate the impact of limiting and overlimiting current modes on the demineralization by ED of a well-characterized whey protein hydrolysate (WPH) solution. Demineralization rate, membrane contact angle and peptide fouling in terms of quantity and sequence identification for both AEMs and CEMs were studied. 2. Materials and methods 2.1. Materials

Fig. 1. Electrodialysis cell configuration for demineralization of the WPH solution. Feed solutions: A: 0.14 M Na2SO4; B: 0.2 M NaCl; C: 0.5% WPH + 0.4% NaCl solution. Solution flow rates were maintained at 100 mL/min.

Food grade Neosepta AMX-SB anion-exchange membranes and CMX-SB cation-exchange membranes were purchased from Astom (Tokyo, Japan). The centrifuge Eppendorf 5804R and a rotor F-346-38 were provided by Eppendorf AG (Hambourg, Germany). The HCl 1.0 N and NaOH 1.0 N were obtained from Fisher Scientific (Nepean, ON, Canada). NaCl and Na2SO4 were supplied by ACP Inc. (Montréal, QC, Canada). Sodium dodecyl sulfate was purchased from Bio-Rad Ltd. (ON, Canada). A BiPRO whey protein isolate (WPI), provided by Davisco Foods International (Eden Prairie, MN, USA) was used for the production of the whey protein hydrolysate (WPH). According to the manufacturer, the WPI composition was 92.7% of protein, 5.0% moisture and 2.3% salt (sodium, potassium, calcium, phosphorus and magnesium). Proteins in WPI consist of 68–75% β-lactoglobulin, 19–25% α-lactalbumin and 2–3% bovine serum albumin. The BiPRO WPI was hydrolyzed with pancreatic bovine trypsin (Reference no. T9201) purchased from Sigma – Aldrich (St. Louis, MO, USA). Then, the hydrolysate was heated at 80 °C during 30 min in order to inactivate the enzyme and avoid further breakdown of peptides [14] and freeze-dried.

in order to remove a potential peptide fouling. The NaCl and SDS desorption solutions were recovered, freeze-dried and analyzed by µBCA protein assay kit for peptide quantification and HPLC-MS for peptide identification. Fouled and desorbed membranes were analyzed in terms of contact angle. Each measurement was performed in triplicate and three independent repetitions were performed for each current density regions. 2.4. Analyses and chemicals 2.4.1. Potential difference, conductivity and pH measurements Current density regions were determined following the Cowan and Brown method [15] in order to estimate the limiting current. Electrical conductivity and pH values of feed and NaCl solutions were measured with a YSI conductivity meter (Model 3100, Yellow Springs,OH) and a pH-meter (SP20 model, ThermoOrion, purchased from VWR International, Montreal, Canada).

2.2. ED cell configuration

2.4.2. Contact angle of membranes Contact angles of fouled and desorbed membranes were measured using a goniometer FTA200 (First Ten Angstroms, Inc., Portsmouth, VA, USA). Excessive water at membrane surface was removed with a filter paper [16] and membrane was immobilized on a platform in a laid flat position. The images of a distilled water drop deposited at the surface of the membrane were captured by a high resolution camera to determine the contact angle value. The measurements were made at room temperature and in triplicate at different spots of the membrane.

A 500 mL solution of 0.5% (w/w) WPH supplemented with 0.4% (w/w) NaCl was demineralized by ED. The ED cell was a MicroFlow type cell from ElectroCell Systems AB (Täby, Sweden) using a Xantrex power supply (HPD 60-5, QC, Canada). The anode, a dimensionallystable electrode (DSA) and the cathode, a 316 stainless steel electrode, were supplied with the cell. Four AMX-SB and five CMX-SB membranes of surface area of 10 cm2 were used. The compartments defined three closed loops containing the feed solution (0.5% WPH), the salt ion recovery (0.20 M NaCl) and the electrode-rinsing solution (0.14 M Na2SO4). Each closed loop was connected to an external plastic reservoir, allowing continuous recirculation at a rate of 100 mL/min for each compartment (Fig. 1).

2.4.3. Peptide concentration in washing solutions The peptide concentrations in the NaCl and SDS washing solutions were solubilized in µBCA™ protein assay kit (Pierce Biotechnology Inc., Rockford, IL, USA). Assays were conducted as recommended by the manufacturer.

2.3. Protocol A 0.5% WPH solution was demineralized during 60 min at different constant voltage corresponding to no current application, underlimiting, limiting and overlimiting modes: the current values for each current modes were determined using current-voltage curve and potential difference measurements (see Sections 2.4.1. and 3.1 for determination methods and values, respectively). According to previous studies, it appeared that both AEMs and CEMs were fouled by peptides at pH 6 [4,5]. Therefore, the pH of the WPH solution was continuously maintained at 6.0 ± 0.5 to avoid a change in peptide charges and to generate a fouling. After ED treatments, AEMs and CEMs were recovered and half of them were soaked separately and successively in 20 mL of 2.0% (w/w) NaCl and in 20 mL of 0.5% (w/w) SDS solutions

2.4.4. Peptide identification in freeze-dried desorption solutions Freezed-dried desorption solutions were solubilized in 1.5 mL of HPLC grade water to optimize peptide concentration for UPLC-MS analyses. RP-UPLC analyses were performed using a 1290 Infinity II UPLC (Agilent Technologies, Santa Clara, CA, USA). The equipment consisted of a binary pump (G7120A), a multisampler (G7167B), an inline degasser and a variable wavelength detector (VWD G7114B) adjusted to 214 nm. Diluted peptides were filtered through 0.22 µm PVDF filter into a glass vial. The sample was loaded (5 µL) onto an Acquity UPLC CSH 130 1.7 µm C18 column (2.1 mm i.d. × 150 mm, Waters Corporation, Milford, MA, USA). The column was operated at a flow 213

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

rate of 400 µL/min at 45 °C. The gradient consisted of solvent A (LC-MS grade water with 0.1% formic acid) and solvent B (LC-MS grade ACN with 0.1% formic acid) starting at 2% B and ramping to 35% B in 40 min, then ramping to 85% B to 40.50 min, holding until 42 min, then back to initial conditions until 45 min. A hybrid ion mobility quadrupole TOF mass spectrometer (6560 high definition mass spectrometry (IM-Q-TOF), Agilent, Santa Clara, USA) was used to identify and quantify the relative abundances of the tryptic peptides. All LC-MS/ MS experiments were acquired using Q-TOF. Signals were recorded in positive mode at Extended Dynamic Range, 2 GHz, 3200m/z with a scan range between 100 and 3200m/z. Nitrogen was used as the drying gas at 13.0 L/min and 150 °C, and as nebulizer gas at 30 psig. The capillary voltage was set at 3500 V, the nozzle voltage at 300 V and the fragmentor at 400 V. The instrument was calibrated using an ESI-L low concentration tuning mix (G1969-85000, Agilent Technologies, Santa Clara, CA, USA). Data acquisition and analysis were carried out using the Agilent Mass Hunter Software package (LC/MS Data Acquisition, Version B.07.00 and Qualitative Analysis for IM-MS, Version B.07.00 with BioConfirm Software).

Fig. 2. Experimental current-voltage curve of the whole electrodialysis cell.

2.4.5. Peptide identification in gel obtained from the WPH using SDS-PAGE SDS-PAGE analyses were performed by using commercially available precast linear gradient polyacrylamide gels in reducing condition (4–20%; Bio-Rad, Missisauga, ON, Canada). Samples were diluted 1:1 (v/v) in a dissociation buffer consisting of 0.5 M Tris–HCl pH 6.8, 0.5% bromophenol blue, 35% glycerol, 5% β-mercaptoethanol, 10% (w/v) SDS solution, and heated in a boiling water for 5 min. Electrophoretic separation was performed in a Multiphor II system applying a maximum voltage of 90 V and a current gradient increasing from 15 to 50 mA with a migration buffer consisting of a 0.02 M Tris (hydroxymethyl) aminomethane, glycine 5 M, SDS (w/v) 0.1% solution. The proteins were stained using standardized protocols (Coomassie blue 0.1%, methanol 50%, acetic acid 10% and water 40%). The gels were destained in a solution containing acetic acid (10%), methanol (40%) and water (50%). Molecular Imager® ChemiDoc™ was used for imaging the gels. SDS-PAGE molecular weight standards (PrecisionPlus, BioRad) were used as protein standards.

and 3.8 ± 0.2d mS/cm, respectively after 60 min at 0.7, 4.5, 9.0 and 45 V. Significant differences were only found at 9.0 and 45 V (probability levels at 9.0 V (p9V) and 45 V (p45V) < 0.001) in comparison with the initial conductivity values. On the opposite, the higher the voltage, the faster the increase of NaCl solution conductivities. Indeed, NaCl conductivity increased from 12.5 ± 0.3 (initially) d to 13.0 ± 0.6 c, d , 13.6 ± 0.3 c, 14.4 ± 0.4 b and 16.0 ± 0.8 a mS/cm, respectively after 60 min at 0.7, 4.5, 9.0 and 45 V. At 0.7 V, no electric current was applied. However, the conductivity of WPH increased very slightly corresponding to a mineralization rate of 7.9 ± 4.5%. Since the salt concentration was higher in the recovery compartment, the sodium and chloride ions could cross the IEMs to migrate into the feed compartment by diffusion. In underlimiting (4.5 V), limiting (9.0 V) and overlimiting (45 V) conditions, the conductivities were significantly lower after 60 min (p ≤ 0.015) and the demineralization rate were 6.2 ± 1.6, 18.1 ± 2.0 and 37.3 ± 4.3%, respectively. These results strengthen previous observations reporting that the mass transfer rate was five times higher using an overlimiting current (equal to three theoretical limiting currents) in comparison with the theoretical value [17]. The pH evolutions of WPH and NaCl solutions were measured for each treatment (Fig. 4). In limiting and overlimiting conditions, the WPH solution was continuously maintained at pH 6.0 ± 0.5 in order to prevent peptide charge modifications due to water splitting. The pH of NaCl solution was initially 5.2 ± 0.1 b and after 60 min, it was

2.4.6. Statistical analyses Data were subjected to one way analyses of variance (ANOVA). Tukey tests were also performed on data using SigmaPlot software (Version 12.0) to determine which treatment was statistically different from the others at a probability level P of 0.05. 3. Results 3.1. Determination of the current density regions For the ED overall potential difference measurements using the Cowan and Brown method, a mathematical transformation curve was plotted in order to determine the theoretical limiting intensity value (Fig. 2). The vertex of obtained parabola-like curve indicates that the limiting intensity value was 69 mA corresponding to a limiting current density (LCD) of 6.9 mA/cm2 and a limiting tension Ulim of 9.0 V. Consequently, the electrode potential difference of the underlimiting, limiting and overlimiting modes have been chosen respectively in this study at 4.5, 9.0 and 45 V. Three additional runs without application of current (residual electrode potential difference of 0.7 V) were carried-out as a control. 3.2. Conductivity and pH measurements Conductivities of WPH and NaCl solutions were measured for each treatment (Fig. 3). Concerning conductivity of the WPH solution, the higher the voltage, the faster the decrease. Indeed, its conductivity varied from 6.2 ± 0.3a,b (initially) to 6.7 ± 0.5a, 5.7 ± 0.5b, 5.1 ± 0.2c

Fig. 3. Electrical conductivity evolution of WPH and NaCl solutions during demineralization.

214

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

5.9 ± 0.1 a, 5.1 ± 0.2 b, 3.8 ± 0.1 c and 3.0 ± 0.2 d at 0.7, 4.5, 9.0 and 45 V, respectively. After treatment, the pH of NaCl solution was significantly higher at 0.7 V, constant at 4.5 V and lower at 9.0 and 45 V, as expected: this confirmed that our underlimiting, limiting and overlimiting regions were well defined. Surprisingly at 45 V, the pH of salt solution increased suddenly to 8.3 ± 1.6 during the two first minutes of the process and then it decreased drastically to 3.0 ± 0.2.

t

3.3. Evaluation of peptide fouling on IEM surfaces 3.3.1. Contact angle of membranes During the ED demineralization step, only the cathode side of AEM and the anode side of CEM were in contact with the WPH solution. Therefore, only these sides are presented on Fig. 5 to be clearer. After ED, AEMs from anode side did not show any significant difference (p > 0.401) comparing to the control ones with a mean value of 61 ± 3°. Concerning AEMs from cathode side (in contact with the WPH solution), contact angles values were significantly lower compared to AEM control (p < 0.001) (Fig. 5a). Indeed, the average value of contact angle was 42 ± 3° at 0.7, 4.5 and 9.0 V whereas it was 53 ± 3° at 45 V. A potential peptide fouling layer by hydrophilic species present on the AEM surface could explain the decrease of contact angle. After soaking of fouled AEMs in both NaCl and SDS desorption solutions (Fig. 5a), the contact angle of the cathode side returned back to its initial value (p > 0.907 comparing to the AEM control). Therefore, fouling peptides were removed from the cathode side of AEMs and were recovered in the desorption solutions. Concerning the CEMs after ED and after soaking in both desorption solutions (Fig. 5b), anode (in contact with the WPH solution) and cathode sides did not present any significant difference comparing to the CEM control (p ≥ 0.992 after ED and p ≥ 0.334 after soaking in both desorption solution) with an average value of 52 ± 3°. Therefore, contact angle did not reveal the presence of peptide on the CEM surface after desorption processes. 3.3.2. Peptide concentrations in desorption solutions Concerning the AEMs, the peptide concentrations in the NaCl desorption solutions were not significantly different at 0.7, 4.5 and 9.0 V (p > 0.756) with a mean value of 13.4 ± 2.1 mg/L whereas at 45 V, it was significantly lower (p < 0.049) with a concentration of 5.1 ± 2.1 mg/L. Peptide concentrations in SDS solutions did not show any significant difference (p > 0.550) with a mean value of 2.8 ± 1.3 mg/L. The peptide concentration was five times lower in SDS than NaCl solutions. For the CEMs, the peptide concentrations in NaCl solutions presented an averaged value of 10.6 ± 0.9 mg/L at 0.7, 4.5 and 9.0 V (p > 0.989) whereas it was significantly lower at 45 V

Fig. 5. Contact angles of (a) cathode side AEMs and (b) anode side CEMs in contact with WPH.

(p < 0.041) with a mean value of 6.8 ± 1.1 mg/L. For all treatment condition of CEMs, the peptide concentration in SDS solutions was of 11.2 ± 1.4 mg/L (p > 0.919) (Fig. 6). According to the peptide concentrations, fouling was globally reduced by 62% and 36% at 45 V for AEM and CEM, respectively compared to the non-overlimiting conditions (0.7, 4.5 and 9.0 V). 3.4. Peptide identification 3.4.1. Peptide profile and characterization in desorption solutions for AEMs In NaCl desorption solutions, the global peptide relative abundances at 0.7, 4.5 and 9.0 V were significantly higher than at 45 V with respective area values of 4.2 ± 1.3; 3.1 ± 1.1; 4.8 ± 1.2 and 1.7 ± 1.4 ×1010AU (Fig. 7). Around 70% of the total abundance were due to four peptides sequences from the β- lactoglobulin as TPEVDDEALEKFDK, VLVLDTDYK, TPEVDDEALEK and IIAEK (Table 1). No peptide was detected in the SDS desorption solutions by MS-MS contrary to µBCA which revealed a low presence of peptides. Since µBCA is a colorimetric technique, it may not be as accurate as the MS-MS. Thereby, peptides involved in the AEM fouling only interacted through electrostatic interactions. 3.4.2. Peptide profile and characterization in desorption solutions for CEMs In NaCl desorption solutions, the global peptide relative abundances at 0.7, 4.5 and 9.0 V were significantly higher than at 45 V with respective area values of 3.6 ± 0.2; 4.1 ± 0.2; 3.8 ± 0.1 and 2.2 ± 1.7 ×1010AU (Fig. 1A, Supplementary material). All identified peptides

Fig. 4. Evolution of the pH of NaCl solution during demineralization.

215

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

4. Discussion 4.1. Hydrolysate demineralization, pH evolution and catalytic effect of peptide ionogenic groups at the membrane interface As expected, the higher the tension, the faster the demineralization of the WPH solution. As observed from these results, a mineralization rate of 7.9 ± 4.5% was obtained at 0.7 V while demineralization rates of 6.2 ± 1.6, 18.1 ± 2.0 and 37.3 ± 4.3% respectively at 4.5, 9 and 45 V. At 0.7 V, the electrical force dragging ions was extremely low and sodium and chloride ions crossed the IEMs to migrate into the feed compartment by diffusion, leading to a mineralization. When a current was applied, the sodium and chloride ions present in the WPH solution (diluate compartment) respectively migrated through the AEMs and CEMs and leaded to a demineralization as a function of the applied voltage. However, these variations in demineralization were also due to the fact that the LCD was reached and water molecules were dissociated into H+ and OH- ions. However, the final demineralization at 45 V was very effective since it was six times and twice higher than at 4.5 and 9 V, respectively. These results strengthen previous observations reporting that the mass transfer rate was five times higher using an overlimiting current (equal to three theoretical limiting currents) in comparison with the theoretical value [17]. In this specific condition of voltage, the intense current leads to chaotic vortices and water splitting in higher proportion. Consequently, water splitting phenomenon induces a decreased in pH of the solution. However, during the experiment in overlimiting conditions, it was observed that the pH of the salt recovery solution increased during the two first minutes of the process and then decreased back. This phenomenon can be explained by the nature of groups at the membrane interface (from either the matrix or peptides) which could impact the dissociation rate of water molecules all along the process (Table 4) [18,19]. Indeed, at the beginning of the process (t = 0), the pH of recovery compartment increased until reaching 8.3 ± 1.6 after only two minutes. Therefore, the water splitting was higher at the AEMs boundary layers than at the CEMs's. However, the AMX-SB membranes intrinsically contain quaternary ammonium groups (-N+(CH3)3) whose nitrogen atom does not carry any lone electron pair and consequently water splitting cannot occur (Table 4: klim = 0 s−1). Nonetheless in overlimiting conditions, these fixed quaternary ammonium groups may be converted into tertiary (=N(CH3)2) or secondary amine (≡N(CH3)) as described with the Hoffmann degradation [20,21]. In addition, Choi et al. noticed this degradation after only one hour under a strong electric field [22]. Those newly formed groups accelerated the water splitting reaction for AEMs (Table 4: klim = 10−1 s−1) such as:

Fig. 6. Peptide concentration in NaCl and SDS desorption solutions after membrane desorption.

came from the β-lactoglobulin. The ALPMHIR and TKIPAVFK sequences represented around 80% of the total abundance (Table 2). Interestingly, the relative abundance of ALPMHIR was higher than TKIPAVFK at 0.7, 4.5 and 9.0 V but lower at 45 V (Fig. 8). In SDS desorption solutions, the global abundances at 0.7, 4.5, 9.0 and 45 V were 0.07 ± 0.09; 0.07 ± 0.01; 0.09 ± 0.00; 0.14 ± 0.03 ×1010AU (data not shown). These values were approximately ten times lower than in NaCl desorption solutions suggesting that hydrophobic interactions have a minor impact on fouling comparing to electrostatic interactions. The sequences involved in the SDS desorption solutions were the same in the salt solutions. Indeed, these sequences include either positive (H, K, R) or hydrophobic (A, L, P, M, I, V) residues able to interact electrostatically and hydrophobically, respectively.

3.4.3. Characterization of a gel present on the surface of AEMs in limiting and overlimiting conditions After ED at 9.0 and 45 V, a white gel was present on some AEMs cathode side. The ratio of membrane surface covered by the gel was measured using the NIH Fiji-Image J (Version 1.48) software (National Institutes of Health, Bethesda, MD,USA). The gel covered 1.7 ± 2.0 and 4.4 ± 2.8% of their total surface at 9.0 and 45 V, respectively (Fig. 9). Consequently, the important decrease of peptide fouling on the AEMs (approximately 60%) is not due to that gel covering a relative small area. Since the WPH is not supposed to contain any protein, a gel formation was not expected on the surface. Since non hydrolyzed proteins might be present into the WPH, a 0.5% (w/w) WPH solution was prepared and adjusted to pH 5.0. At this pH, a white flocculation appeared. Therefore, the solution was centrifuged at 5000 rpm during 10 min and a precipitate was observed after centrifugation. That precipitate and the white gel on the AEMs at 9.0 and 45 V were characterized using SDS-PAGE and RPUPLC-MS-QTOF analyses. The SDS-PAGE (Fig. 10) revealed the presence of β-lactoglobulin and α-lactalbumin in the initial 0.5% (w/w) WPH solution. After centrifugation, both proteins partially moved into the precipitate phase demonstrating that they were partially denatured by a thermic pre-treatment of WPI and WPH. Besides, every molecules whose MW < 10 kDa moved into the precipitate. It appeared that these "small" molecules were similar to those present in the white gel after ED. The MSQTOF analyses (Fig. 2A, Supplementary material) showed profiles including 18 peaks. The "small" molecules visualized in the SDS-PAGE may be the macropeptides whose MW are between 2 and 4 kDa (Table 3). It is possible that these macropeptides formed a gel due to their sizes and structures. Moreover, other peptide sequences already detected in the salt solutions for both membranes were identified in the gels. These peptides might have interacted with the macropeptides during the gelation step. All peptides peptides were originating from β-lactoglobulin.

+ − ⎧ B + H —OH ↔ BH + OH + + H O ↔ B + H O+ BH 2 3 ⎨ ⎩

with B, a base carrying at least a lone electron pair such as =N(CH3)2 In contrast for CEMs, that water dissociation was slower in presence of their fixed sulfonic groups (Table 4: klim = 3·10−3 s−1). Indeed, Simons revealed that water splitting was impossible/very slow in pure NaCl solutions but could be generated in presence of weakly acidic groups such as carboxylic acid (–COO-) available in amino acid residues [23–26] as explained by the Second Wien Effect [27]. Since water splitting for CEMs was negligible at t = 0, it is possible that only sodium ions were present at their boundary layers due to repulsions between the peptides and the fixed sulfonic groups of CEMs. However after two minutes, the pH of recovery compartment started to decrease back meaning that the water dissociation became greater for the H+ generated by CEMs than for the OH- generated by AEMs. Indeed, under a strong electrical field, the positively charged peptides (such as ALPMHIR and TKIPAVFK) were quickly attracted by the cathode and moved into the boundary layer of the CEMs. Consequently, the catalytic carboxylic acid at their C-terminal enhanced the generation of water 216

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

Fig. 7. RP-UPLC-MS-MS profiles of salt desorption solutions after peptide desorption from fouled AEMs at (a) 0.7, (b) 4.5, (c) 9.0 and (d) 45 V.

splitting (Table 4: klim = 10 s−1). At the same time, negative peptides (such as VLVLDTDYK and TPEVDDEALEK) were attracted by the anode and moved into the boundary layer of the AEMs. However, according to the Second Wien Effect [27], weak bases such as the ammonia group (–NH3+) at their N-terminal could not affect catalytically the dissociation of water molecules. Consequently, the apparent pH in the salt recovery compartment decreased to 3.0 ± 0.2 while it tended to increase in the WPH compartment but was maintained at 6.0 by adding HCl to neutralize the hydroxyl ions. Nonetheless, those added protons were able to cross the CEMs into the salt recovery compartment and consequently, were partially responsible for the pH decrease. In overlimiting conditions, the first addition of HCl was around eight min (t = 8) which explains the shoulder observed in the pH curve at that moment (Fig. 4). Consequently, this addition of HCl did not affect the decrease of pH at two minutes (t = 2) but only the final pH value which

was underestimated due to water splitting. 4.2. Peptide fouling At pH 6, most of the AEM peptide sequences fouling the AEMs was globally negative whereas all peptides found on the CEMs were positive (Table 5). Consequently, the negative D and E residues (pKa = 4.0) could interact with the positive groups of the AEMs (–N+(CH3)3). Besides, the positive K (pKa = 10.5) and R (pKa = 12.5) residues interacted with the negative groups of the CEMs (SO3-). Thereby, these results demonstrated that fouling of IEMs during ED treatment was mainly due to electrostatic peptide/membrane interactions as described in previous studies with no current application [4,5]. However, when comparing the present results in ED conditions with those from previous studies, it appeared that current application did not seem to 217

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

Table 1 Identification of peptides detected in salt and SDS solutions after desorption from fouled AEMs. Peak number

Salt desorption solutions 1 2 3 4 5 6 7 8 9 10 11 SDS desorption solutions No peak detected

Retention time (min)

6.26 8.61 11.10 11.81 12.38 14.31 15.75 16.18 16.39 17.94 25.56

Peptide sequence

Molecular weight (Da)

IDALNENK Not identified IIAEK TPEVDDEALEK VLVLDTDYKK Not identified Not identified TPEVDDEALEKFDK VLVLDTDYK Not identified VYVEELKPTPEGLEILLQK

Relative abundance (%)

Theoretical

Measured

915.47 – 572.35 1244.58 1192.67 – 1590.74 1634.77 1064.58 – 2312.25

915.4781 1046.49 572.3573 1244.5951 1192.6893 1418.6801 1590.7723 1634.7945 1064.5899 442.3183 2312.2921

5.6 ± 3.9 2.9 ± 1.1 11.0 ± 3.7 14.5 ± 4.2 0.3 ± 0.2 5.5 ± 1.5 12.9 ± 5.5 22.2 ± 3.5 20.2 ± 2.7 2.9 ± 2.3 1.9 ± 1.2

Table 2 Identification of peptides detected in salt and SDS solutions after desorption from fouled CEMs. Peak number

Salt desorption solutions 1 2 3 4 5 SDS desorption solutions 1 2 3 4

Retention time (min)

Peptide sequence

Molecular weight (Da) Theoretical

Measured

relative abundance (%)

3.57 8.87 10.96 12.38 16.19

LIVTQTMK ALPMHIR TKIPAVFK IPAVFK VLVLDTDYK

932.54 936.47 902.56 673.42 1064.58

932.5501 836.4797 902.5718 673.4237 1064.5751

10.3 ± 1.9 43.8 ± 12.5 37.5 ± 14.5 8.6 ± 2.7 1.3 ± 0.9

3.59 8.92 11.18 12.43

LIVTQTMK ALPMHIR TKIPAVFK IPAVFK

932.54 936.47 902.56 673.42

932.5360 836.4693 902.5592 673.4162

15.2 ± 9.7 56.9 ± 5.3 23.9 ± 10.9 4.0 ± 1.3

Fig. 9. White gel on the cathode side of AEMs at (a) 9.0 V and (b) 45 V.

interact electrostatically with the membrane through their positive residues (first layer) and then hydrophobically between themselves (second layer) in a disordered way. Finally, some positive residues may be located at the top of the second layer and may be able to interact with the water drop during the contact angle measurement. On the contrary, under a current field, each positive residues would be attracted by the cathode. Therefore, they would be covered by the hydrophobic residues and would be located underneath. Consequently, during goniometry, the water drop would interact more with those charged groups of hydrophobic residues than with the positive ones: the contact angle remained stable despite the presence of peptides. A peptide-made gel was detected on the cathode side of AEMs when water splitting occurred (9.0 and 45 V) (Table 3). Usually, peptides

Fig. 8. Relative abundances of peptides responsible for fouling of CEMs.

change drastically the peptide fouling profiles since the peptide sequences were the same respectively for the AEMs and CEMs. Nonetheless, the application of current may have affected the peptide orientation towards the membrane surface. Indeed, the contact angle of fouled CEM did not show any significant difference with the CEM control. However, in a recent study, the CEMs fouled with the same WPH in static conditions presented lower contact angles [5]. The fouled peptides were composed of either positive (H, K, R) or hydrophobic (A, L, P, M, I, V, F) residues. In static conditions, those peptides would 218

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

4.3. Impact of overlimiting conditions and membrane nature on peptide fouling According to the literature, water splitting would create a protonbarrier able to hamper the mineral scaling growth [30]. Considering the peptide concentrations in desorption solutions and RP-UPLC-MS-MS analysis, water splitting occurring in limiting conditions (9.0 V) did not affect the peptide fouling despite its "barrier effect". Therefore, the H+ and OH- generated at the interface of IEMs were not sufficient to modify the global charges of peptides. Nonetheless, it is worth to note that 1.0 ± 0.4 and 7.5 ± 1.4 mL of a 0.5 N HCl solution were added in the WPH compartment in order to maintain its pH at 6.0 ± 0.5 at 9.0 and 45 V, respectively: the water splitting rate was higher in overlimiting than limiting conditions. Moreover in overlimiting conditions (45 V), fouling was approximately reduced by 60% and 38% for AEMs and CEMs, respectively. This decrease in peptide fouling cannot just be due to water splitting since overlimiting conditions were more effective although the water splitting phenomenon was more important. Another phenomenon such as electroconvective vortices could explain this high decrease in peptide fouling. Theoretically, vortices enhance the salt counterion transport through the IEMs by current induced convection in the depletion zone near the IEM interface. Therefore in overlimiting conditions, it appeared that (1) water splitting was important enough to hamper the peptide fouling through its "barrier effect" and (2) electroconvective vortices could prevent partially the peptide fouling on the membranes by scrubbing their surfaces and increasing the turbulence into the compartments. Another interesting information concerns the charge distribution of peptides fouled on CEMs. Indeed, ALPMHIR interacted 68% less while TKIPAVFK interacted 25% more in overlimiting conditions in terms of mass balance. The positive charge distribution over the peptides surface may be an explanation (Fig. 11). Indeed in the bulk solution, ALPMHIR carries two positive charges far away from each other: the N-terminal (pKa = 9.8) and the side-chain of R7 (pKa = 12.5). At the CEM interface, hydroxyl ions are generated (pH > 6) and consequently, the Nterminal is getting neutral and non-attractive to the sulfonic group of the CEM (SO3-). The side-chain of R7 may be still positive but since it is spatially close to its negative C-terminal (pKa = 2.1), they would neutralize each other preventing partially interactions between R7 and the sulfonic groups. On the contrary, the TKIPAVFK sequence carries three positive charges uniformly distributed on the same plane which is completely recovered: the N-terminal (pKa = 9.8) and the two side-

Fig. 10. SDS-PAGE profile of (Std) MW standard, (DStd) MW dairy standard, (1) WPH before centrifugation, (2) supernatant, (3) precipitate and (4) gel on the AEM after electrodialysis.

cannot precipitate because of their relative low MW. However, many macropeptides (MW between 2 and 4 kDa) were detected suggesting that the gel formation would be possible. In the ED cell (Fig. 1), the WPH compartment was located between the anode side of CEM (generating hydroxyl ions) and the cathode side of AEMs (generating protons). Therefore, in the boundary layer of AEM, the pH was much lower than 6. When coming into or close to that boundary layer, the macropeptides precipitated on the membrane surface (pI=pH) forming a chaotic network and a gel. Indeed, this was confirmed by the appearance of a white flocculation in the WPH when pH was adjusted close to the isoelectric points of the main macropeptides (Table 5). That observation confirms a previous study dealing with the demineralization of whey proteins which led to a visible heterogeneous deposit in acidic and neutral pH conditions on the AEM interface when H+ leakage occured [28]. Moreover, another study demonstrated the gel formation by low molecular-weight peptides from β-lactoglobulin at a pH close to their pI [29].

Table 3 Identification of peptides detected in the white gel on the cathode side of AEMs and in the white precipitate after centrifugation. Peak number

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Retention time (min)

5.94 6.22 10.90 11.81 12.38 16.19 16.41 17.67 25.49 26.33 26.66 28.54 28.87 29.32 29.78 30.68 35.26

Peptide sequence

Molecular weight (Da)

GLDIQK IDALNENK LIVTQTMK TPEVDDEALEK VLVLDTDYKK TPEVDDEALEKFDK VLVLDTDYK IDALNENKVLVLDTDYKK VYVEELKPTPEGDLEILLQK VYVEELKPTPEGDLEILLQKWENGECAQKK Not identified Not identified VYVEELKPTPEGDLEILLQKWENGECAQK Not identified YLLFCMENSAEPEQSLACQCLVR Not identified VAGTWYSLAMAASDILLDAQSAPLR

219

% abundance

Theoretical

Measured

672.38 915.47 932.54 1244.58 1192.67 1634.77 1064.58 2090.13 2312.25 3485.78 – – 3357.68 – 2646.20 – 2619.34

672.3813 915.4662 932.5366 1244.5778 1162.6711 1634.7674 1064.5748 2090.1264 2312.2497 3485.7757 3312.5642 1792.8074 3357.6772 2161.9519 2646.1797 2676.1478 2619.3673

0.8 ± 0.8 2.8 ± 0.4 5.0 ± 0.6 0.9 ± 0.2 4.2 ± 0.7 6.3 ± 0.2 7.1 ± 0.3 0.6 ± 0.4 31.4 ± 9.0 8.7 ± 0.1 1.3 ± 0.3 8.5 ± 0.5 2.0 ± 0.1 3.0 ± 0.1 7.0 ± 0.5 1.7 ± 0.0 5.4 ± 1.7

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

chains of K2 and K8 (pKa = 10.5). When considering the CEM interface (pH > 6), some of those sites may be neutral but remaining positive charges among the three of them may still exist and able to interact with the sulfonic groups.

and 38% at 45 V for AEM and CEM, respectively. Nonetheless, recent studies reported that water splitting and electroconvective vortices would deteriorate IEMs for a long duration exposures [17,31]. In addition, the energy consumption was important because of intensive voltages (45 V). Thereby, it is necessary to apply a lower voltage but still in overlimiting conditions (between 9.0 and 45 V) in order to preserve the vortices and water splitting hampering the fouling and enhancing the demineralization at a lesser cost. More interestingly, recent practices of ED highlighted the advantages of non-stationary currents such as the pulsed electrical field which would be able to prevent membrane scaling and fouling while consuming less energy and preserving the integrity of IEMs [32].

5. Conclusion

Acknowledgements

It appeared from these results that fouling of IEMs during ED demineralization treatment of hydrolysate was mainly due to electrostatic interactions between IEMs and oppositely charged peptides. However, the peptide sequences and quantity were similar when applying relative low voltages (0.7, 4.5, 9.0 V) but lower in overlimiting voltage (45 V). To the best of our knowledge, this article showed for the first time the positive influence of the overlimiting conditions on the peptide fouling according to the IEM nature: the fouling was globally reduced by 60%

The financial support of the Natural Sciences and Engineering Research Council of Canada (NSERC) Grant no. 210829409 is acknowledged. This work was realized within the French-Canadian International Associated Laboratory on “Bioproduction of natural antimicrobials” (LIAAN). The authors thank Diane GAGNON and Jacinthe THIBODEAU, research professionals at Université Laval, for their great implication into this project.

Table 4 Ionogenic groups of membranes arranged in the increasing order of rate constants. for the dissociation of water (adapted from Zabolotskii et al. [19]). Groups

–N+ (CH3)3

–SO3−

=NH/-NH2

–COO−

klim (s−1)

0

3·10−3

10−1

10

Table 5 Physicochemical characteristics of whey tryptic peptides recovered in the desorption solutions and gel after demineralization at 9.0 and 45 V or precipitated after acidification and centrifugation. Anion-exchange membrane

Cation-exchange membrane

Gel/precipitate

Peptide sequence

Global charge

Peptide sequence

Global charge

Peptide sequence (MW > 2 kDa)

pI

IDALNENK IIAEK TPEVDDEALEK VLVLDTDYKK TPEVDDEALEKFDK VLVLDTDYK VYVEELKPTPEGLEILLQK

− 0 − 0 − − −

LIVTQTMK ALPMHIR TKIPAVFK IPAVFK

1 2 2 1

IDALNENKVLVLDTDYKK VYVEELKPTPEGDLEILLQK VYVEELKPTPEGDLEILLQKWENGECAQKK VYVEELKPTPEGDLEILLQKWENGECAQK YLLFCMENSAEPEQSLACQCLVR VAGTWYSLAMAASDILLDAQSAPLR

4.7 4.3 4.6 4.3 4.3 4.2

1 4 4 1 2

Fig. 11. Electrostatic potential of the ALPMHIR (left) and TKIPAVFK (right) sequences in the bulk WPH solution at pH 6 (up) and at the CEM interface in limiting and overlimiting conditions (down) (adapted from Persico et al. [5]). Blue, red and white-colored spots indicate peptide regions with positive, negative and neutral electrostatic potential, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

220

Journal of Membrane Science 543 (2017) 212–221

M. Persico et al.

5330–5340. [15] D.A. Cowan, J.H. Brown, Effect of turbulence on limiting current in electrodialysis cells, Ind. Eng. Chem. 51 (12) (1959) 1445–1448. [16] C. He, et al., Surface orientation of hydrophilic groups in sulfonated poly(ether ether ketone) membranes, J. Colloid Interface Sci. 409 (0) (2013) 193–203. [17] N.D. Pismenskaya, et al., Evolution with time of hydrophobicity and microrelief of a cation-exchange membrane surface and its impact on overlimiting mass transfer, J. Phys. Chem. B 116 (7) (2011) 2145–2161. [18] V.I. Zabolotskii, M.V. Sharafan, N.V. Shel’deshov, Influence of the nature of membrane ionogenic groups on water dissociation and electrolyte ion transport: a rotating membrane disk study, Russ. J. Electrochem. 44 (10) (2008) 1127–1134. [19] V.I. Zabolotskii, N.V. Shel'deshov, N.P. Gnusin, Dissociation of water molecules in systems with ion-exchange membranes, Russ. Chem. Rev. 57 (8) (1988) 801. [20] M.-S. Kang, Y.-J. Choi, S.-H. Moon, Characterization of anion-exchange membranes containing pyridinium groups, AIChE J. 49 (12) (2003) 3213–3220. [21] V.I. Zabolotskii, R.K. Chermit, M.V. Sharafan, Mass transfer mechanism and chemical stability of strongly basic anion-exchange membranes under overlimiting current conditions, Russ. J. Electrochem. 50 (1) (2014) 38–45. [22] J.-H. Choi, S.-H. Moon, Structural change of ion-exchange membrane surfaces under high electric fields and its effects on membrane properties, J. Colloid Interface Sci. 265 (1) (2003) 93–100. [23] R. Simons, Strong electric field effects on proton transfer between membrane-bound amines and water, Nature 280 (5725) (1979) 824–826. [24] R. Simons, The origin and elimination of water splitting in ion exchange membranes during water demineralisation by electrodialysis, Desalination 28 (1) (1979) 41–42. [25] R. Simons, Electric field effects on proton transfer between ionizable groups and water in ion exchange membranes, Electrochim. Acta 29 (2) (1984) 151–158. [26] R. Simons, Water splitting in ion exchange membranes, Electrochim. Acta 30 (3) (1985) 275–282. [27] L. Onsager, Deviations from Ohm's Law in weak electrolytes, J. Chem. Phys. 2 (9) (1934) 599–615. [28] E. Ayala-Bribiesca, et al., Effect of concentrate solution pH and mineral composition of a whey protein diluate solution on membrane fouling formation during conventional electrodialysis, J. Membr. Sci. 280 (1–2) (2006) 790–801. [29] D. Doucet, E.A. Foegeding, Gel formation of peptides produced by extensive enzymatic hydrolysis of β-Lactoglobulin, Biomacromolecules 6 (2) (2005) 1140–1148. [30] N. Cifuentes-Araya, G. Pourcelly, L. Bazinet, Water splitting proton-barriers for mineral membrane fouling control and their optimization by accurate pulsed modes of electrodialysis, J. Membr. Sci. 447 (0) (2013) 433–441. [31] N.D. Pismenskaya, et al., Growth in the velocity of mass transfer through the CMX membrane during its aging by operation in intense current regimes, Pet. Chem. 51 (8) (2011) 610–619. [32] S. Mikhaylin, et al., How physico-chemical and surface properties of cation-exchange membrane affect membrane scaling and electroconvective vortices: influence on performance of electrodialysis with pulsed electric field, Desalination 393 (2016) 102–114.

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.memsci.2017.08.039. References [1] K.V. Kotsanopoulos, I.S. Arvanitoyannis, Membrane processing technology in the food industry: food processing, wastewater treatment, and effects on physical, microbiological, organoleptic, and nutritional properties of foods, Crit. Rev. Food Sci. Nutr. 55 (9) (2015) 1147–1175. [2] O. Roczanski, V. Romanovicz, D.E.O.S. Carpenter, Application of electrodialysis technique to recover contaminated rinsing water from gold electroplating processes, Trans. IMF 90 (3) (2012) 161–168. [3] C. Huang, et al., Application of electrodialysis to the production of organic acids: state-of-the-art and recent developments, J. Membr. Sci. 288 (1–2) (2007) 1–12. [4] M. Persico, et al., How peptide physicochemical and structural characteristics affect anion-exchange membranes fouling by a tryptic whey protein hydrolysate, J. Membr. Sci. 520 (2016) 914–923. [5] M. Persico, et al., Formation of peptide layers and adsorption mechanisms on a negatively charged membrane surface, Journal of Colloid and Interface Science (2017). [6] H. Strathmann, Electrodialysis, a mature technology with a multitude of new applications, Desalination 264 (3) (2010) 268–288. [7] V.V. Nikonenko, et al., Intensive current transfer in membrane systems: modelling, mechanisms and application in electrodialysis, Adv. Colloid Interface Sci. 160 (1–2) (2010) 101–123. [8] Y. Tanaka, Water dissociation reaction generated in an ion exchange membrane, J. Membr. Sci. 350 (1–2) (2010) 347–360. [9] A. Bukhovets, et al., The influence of current density on the electrochemical properties of anion-exchange membranes in electrodialysis of phenylalanine solution, Electrochim. Acta 56 (27) (2011) 10283–10287. [10] E.D. Belashova, et al., Overlimiting mass transfer through cation-exchange membranes modified by Nafion film and carbon nanotubes, Electrochim. Acta 59 (0) (2012) 412–423. [11] I. Rubinstein, B. Zaltzman, Electro-osmotically induced convection at a permselective membrane, Phys. Rev. E 62 (2) (2000) 2238–2251. [12] S. Mikhaylin, et al., Intensification of demineralization process and decrease in scaling by application of pulsed electric field with short pulse/pause conditions, J. Membr. Sci. 468 (2014) 389–399. [13] V.V. Nikonenko, et al., Desalination at overlimiting currents: state-of-the-art and perspectives, Desalination 342 (2014) 85–106. [14] S. Suwal, E. Rozoy, M. Seifu Manenda, A. Doyen, L. Bazinet, Comparative study of in-situ and ex-situ enzymatic hydrolysis of milk protein and separation of bioactive peptides in an electromembrane reactor, ACS Sustain. Chem. Eng. 5 (6) (2017)

221