Interactions of Phenolic Acids and -Glucan: Studies of Adsorption Isotherms and Thermodynamics

Interactions between polyphenols and dietary fiber might play an important role in polyphenol bioactivities. These interactions can be studied through adsorption processes. The aim of this study was to investigate the adsorption of phenolic acids (p-coumaric acid, caffeic acid, chlorogenic acid) onto dietary fiber – β-glucan. Adsorption was carried out at different temperatures (25, 37, and 45 °C) and pH values (1.5, 5.5, and 10). Non-linear isotherm adsorption models (Langmuir, Freundlich, Dubinin-Radushkevich, Temkin, Hill) were applied to analyze the data. Experimentally determined adsorption capacities showed some fluctuations with temperature and pH. Adsorption isotherms modelled the experimentally determined adsorption capacities well. According to isotherm parameters, it can be suggested that the adsorption was a physical process with non-covalent bonding at all temperatures and pH. Thermodynamic parameters showed that the adsorption was spontaneous (except for chlorogenic acid) and exothermic. Adsorption isotherms and thermodynamics gave useful information about phenolic acid – b-glucan interactions.


Introduction
Polyphenols are secondary plant metabolites that have shown numerous positive bioactivities [1][2][3][4] . More recent studies have also shown that they can interact with macromolecules like proteins, carbohydrates, and lipids in the digestive tract, which can affect their bioactivities [5][6][7] . Namely, interactions with other compounds in the digestive tract can affect polyphenol accessibility for absorption (bioaccessibility), the actual amount that is being absorbed (bioavailability), and therefore their bioactivities 5 . For example, dietary fibers can interact with polyphenols, "carry" them to the lower parts of the digestive tract 8 and by doing so, possibly influence their bioactivities. That is why interactions between dietary fibers and polyphenols are in the focus of present studies.
To obtain more information about polyphenol and dietary fiber interactions, the adsorption process between polyphenols and dietary fibers has to be studied. Adsorption is a process in which molecules from the solution (adsorbate) adsorb onto the surface of the adsorbent 9 . The explanation of the adsorption process includes various aspects of the transportation of adsorbate, like transport of the adsorbate from the solution to the adsorbent surface, adsorbate diffusion across adsorbent liquid film, intraparticle diffusion, and adsorption and desorption from the surface of the adsorbent 10 . After reaching the adsorption equilibrium, data about the adsorption process can be obtained (the amount of adsorbate adsorbed onto the adsorbent and the concentration that remains unadsorbed in the solution at constant temperature), which can be modelled with equations called adsorption isotherms 9,11,12 . Some frequently used adsorption isotherm models are Langmuir, Freundlich, Dubinin-Radushkevich, Temkin, and Hill models. These models can provide various information about the adsorption process, such as the physical or chemical nature of the adsorption, or the maximal predicted, theoretical adsorption capacity 9,11 .
One of the natural dietary fibers which can be found in cereals, mushrooms, seaweed, and yeast is β-glucan 13 . Earlier studies have shown that β-glucan can interact with polyphenols [14][15][16][17][18] . Interactions be-tween various polyphenols and β-glucan 17 , tea polyphenols and β-glucan 14,18 , quercetin and β-glucan 16 , vanillin-inspired phenolic derivatives and β-glucan 15 have been investigated. Only some of these studies have analyzed the experimental data with the use of adsorption isotherms 14,18 . However, interactions between polyphenols and β-glucan are yet to be completely understood. They can be affected by environmental conditions, like pH and temperature or by the chemical structure of polyphenols or β-glucan. In order to understand how polyphenols bond to β-glucan, obtain the adsorption capacity, and to understand how environmental conditions affect interactions, further studies are necessary. Moreover, by studying polyphenol -β-glucan interactions, additional insight into dietary fiber -polyphenol interactions in general can be obtained.
The objective of the present research was to study interactions between β-glucan and polyphenols that belong to the phenolic acid subclass, through the adsorption process. To understand the influence of the environment on the adsorption process, the adsorptions were conducted at three different temperatures and at three pH values. Results were analyzed with non-linear isotherm models (Langmuir, Freundlich, Dubinin-Radushkevich, Temkin, and Hill) to help understand the adsorption process. Prediction of the adsorption process that is more precise can be obtained by applying non-linear models, rather than linear models, to the experimental results 19 .
The thermodynamic parameters of adsorption were calculated (standard reaction Gibbs energy, standard reaction enthalpy, and standard reaction entropy).

Adsorption between phenolic acids and -glucan
Adsorption process was performed between phenolic acids (p-coumaric acid, caffeic acid and chlorogenic acid) ( Fig. 1) and barley β-d-glucan in single solution. Adsorptions between phenolic acids and β-glucan were carried out according to procedure from earlier study with some modifications 17 . The model solution in a plastic tube contained β-glucan as the adsorbent (5 mg L -1 ), a phenolic acid as the adsorbate (25 mg L -1 ), and buffer solution. The total volume of the model solution was 500 µL. Adsorption was performed for 16 h. Concentrations of phenolic acids and β-glucan were 25 and 5 mg L -1 , respectively. Adsorptions were conducted at three different temperatures (25 °C, 37 °C and 45 °C, pH adjusted at 5.5). Temperature of 37 °C was chosen since it is a human body temperature. Adsorption at two additional temperatures 25, and 45 °C, enabled us to calculate thermodynamic parameters. To cover higher pH range, adsorption was conducted at pH 1.5, 5.5 and 10, with temperature adjusted to 37 °C. Phenolic acid concentrations in plastic tubes were initial concentrations (c initial , mg L -1 ). Solutions were vortexed (Grant Bio, Cambrid geshire, England) and put in an incubator (Incubator IN 30, Memmert, Schwabach, Germany). After reaching equilibrium (obtained experimentally, 16 h), model solutions were centrifuged (Eppendorf Minispin centrifuge, Eppendorf, Hamburg, Germany) through polyethersulfon membrane (Vivaspin 500, F i g . 1 -Chemical structure of phenolic acids 100-500 µL Sartorius, Goettingen, Germany). In the solution that passed on the other side of membrane, phenolic acids were determined using the spectrophotometric Folin-Ciocalteu method (c, mg L -1 ). The same adsorption with the same conditions was conducted without β-glucan, which represented a blank experiment. Phenolic acids that passed through the membrane in the blank experiment, were determined using spectrophotometric Folin-Ciocalteu method validated in our laboratory for each compound in single solution 20 (c 0 , mg L -1 ). The concentration of adsorbed phenolic acids (c adsorbed , mg L -1 ) was calculated as: The concentration of unadsorbed phenolic acid at equilibrium (c e , mg L -1 ) was calculated as: The q e (amount of adsorbed phenolic acids in milligrams per gram of β-glucan) was calculated according to eq. (3): where c adsorbed is the concentration of adsorbed phenolic acids (mg L -1 ), V m is the total volume of a model solution (L), γ a is the β-glucan concentration (g L -1 ), and V a is the volume of model solution (L). The same adsorption experiment was repeated with various initial phenolic acid concentrations (25, 50, 75, 100, and 150 mg L -1 ) and β-glucan (5 mg L -1 ).

Spectrophotometric Folin-Ciocalteu method
The phenolic acids in the adsorption experiment were determined using the Folin -Ciocalteu method 21 according to our earlier developed and validated procedure 20 . Aliquots of 20 μL of solution after adsorption, 1580 μL of distilled water, 100 μL of Folin -Ciocalteu reagent, and 300 μL of Na 2 CO 3 (200 g L -1 ) were added into a glass tube. The resulting solution was mixed in the vortex (Grant Bio, Cambridgeshire, England), incubated at 40 °C for 30 min (incubator Memmert IN 30, Schwabach, Germany), and then analyzed spectrophotometrically at 765 nm (UV-Vis spectrophotometer UV 2005, Selecta, Barcelona, Spain). A blank solution containing 20 µL of distilled water instead of solution after adsorption was used during the analysis. Calibration curves were constructed for each individual phenolic acid by using the same method, in the concentration range from 1 to 500 mg L -1 . The concentrations of phenolic acids after adsorption were determined (mg L -1 ) using the constructed calibration curves.
The q e is the amount of phenolic acids adsorbed per gram of β-glucan at equilibrium state (mg g -1 ), c e is the phenolic acid concentration in the solution at equilibrium (mg L -1 ), K L is the Langmuir equilibrium constant of adsorption (L mg -1 ) or apparent affinity constant, q m is the theoretical maximum adsorption capacity of β-glucan (mg g -1 ), K F is the Freundlich constant indicative of the relative adsorption capacity of β-glucan ((mg g -1 ) (L mg -1 ) 1/n )), 1/n is the intensity of adsorption, q s is the theoretical isotherm saturation capacity (mg g -1 ), β is a constant related to the adsorption capacity (mol 2 J -2 ), ε is the Polanyi potential (J mol -1 ), R is the gas constant (8.314 J mol -1 K -1 ), T is the temperature (K), E is the adsorption mean free energy (J mol -1 ), c s is the theoretical saturation concentration or solubility (mg L -1 ), A is the equilibrium binding constant related to the maximum binding energy (L g -1 ), apparent Temkin adsorption potential, b T is the heat of adsorption (J mol -1 ), n H is the Hill cooperativity coefficient , and K D is the Hill constant (mg L -1 ).
Langmuir, Freundlich and Temkin equations are two-parameter equations and nonlinear modeling was used to determine parameters K L and q m from Langmuir, K F and 1/n from Freundlich, and b T and A from Temkin equation. Dubinin-Radushkevich isotherm is a three-parameter model in which q s , β and c s can be determined with nonlinear modeling. The c s (mg L -1 ) represents a saturation concentration, a concentration of non-adsorbed phenolic acids in a theoretically saturated solution. Our experiment was not constructed to measure in such large theoretical concentrations, so the c s values were predetermined according to our data for c e . Due to predetermined c s , Dubinin Radushkevich isotherm was used to determine two parameters, q s and β. The β was finally used to calculate mean free energy of adsorption E (eq. 8). Finally, two parameters were reported, q e and E. The Hill equation was finally used to determine parameters q m , n H and K D .

Thermodynamic parameters
The standard reaction Gibbs energy (J mol -1 ) was calculated according to eq. 12: where R is the gas constant (8.314 J mol -1 K -1 ), T is absolute temperature, and K L is the Langmuir equilibrium constant of adsorption for neutral adsorbates or adsorbates with very weak charges 25 expressed in L mol -1 . The Langmuir constant K L was recalculated to a dimensionless parameter 26 . Standard reaction enthalpy and standard reaction entropy were calculated from linear Van't Hoff equations 23,24 (eq. 13), which represents lnK L vs 1/T: Standard reaction enthalpy ΔH 0 was calculated from the slope (-ΔH 0 /R), and standard reaction entropy ΔS 0 from the intercept (ΔS 0 /R).

Statistical analysis
All adsorption experiments were conducted in two parallels, each concentration was measured three times (n = 6). All calculations were performed in MS Excel (Redmond, USA) software. Tool Solver in MS Excel was selected for non-linear regression analysis of adsorption isotherms. The sum of squared errors (SSE) of nonlinear least square regression was calculated according to eq. 14, where c e,i and q e,i are the means of the measured c e and q e values, f (c e , a, b) is the nonlinear model function with generic parameters a and b, and n i is number of data points for the i th initial concentration. Standard error of nonlinear regression was calculated according to eq. 15. (where N is the total number of observations, and k = 1, 2 or 3 is the number of parameters. For statistical comparisons, a one-way analysis of variance with post hoc Tukey test was performed in Minitab (Pennsylvania, USA).

Results and discussion
Adsorption of phenolic acids onto b-glucan (influence of temperature and pH) Table 1 shows the amount of phenolic acid adsorbed per mass of β-glucan (q e , mg g -1 ) determined in the experiment with initial phenolic acid concentration of 25 mg L -1 , at three different temperatures and three different pH values. The q e values were in the range of 48 -438 mg g -1 . It was difficult to compare these results with literature, since β-glucan can vary depending on the source (oats or barley), molecular weight 27,28 , or even environmental factors during growth of barley 25 , which can lead to great difference in adsorption. In addition, the ratio of polyphenol/β-glucan can affect q e 14 . Thus, earlier studies have reported various values for q e in adsorption onto oats β-glucan. Namely, standards of Ta b l e 1 -Amount of phenolic acids adsorbed per g of β-glucan (q e , mg g -1 ) at different temperatures (25, 37, 45 °C (pH 5.5)) and pH values (pH 1.5, 5.5, 10 (37 °C)) Polyphenol q e * (mg g -1 )

°C 37 °C 45 °C
Tukey test** *different letters in a column correspond to differences between polyphenols at the same temperature or pH (obtained with posthoc Tukey test) **different letters in a row correspond to differences between temperature or pH for each polyphenolic compound separately (obtained with post-hoc Tukey test) polyphenols found in tea adsorbed onto oat β-glucan in the amount of 156 to 405 mg g -1 14 . Tea polyphenols adsorbed onto b-glucan at different pH values up to 116 mg g -1 18 . Some polyphenols adsorbed onto plant cell walls up to 600 mg g -1 29,30 , 1200 mg g -1 7 , 1000 mg g - 1 31 . The adsorption in these studies is similar to the results for adsorption of phenolic acids onto β-glucan in this study. Furthermore, the q e for each phenolic acid showed some significant fluctuations with temperature (Table 1). This was similar to earlier studies that had shown that temperature influences the adsorption of tea polyphenols 18 or (-)-epigallocatechin-3-gallate 14 onto oat β-glucan or the adsorption of procyanidins onto apple cell wall material 32 . Furthermore, some fluctuations in q e with pH from 1.5 to 10 could also be seen for each compound (Table  1). This was similar to earlier studies that had shown that pH can influence the adsorption of (-)-epigallocatechin-3-gallate onto β-glucan in the range of pH 2 -7 14 or tea polyphenols onto β-glucan in the range from pH 3 to 7 18 . Phenolic acids like caffeic and chlorogenic acid also showed pH dependence in the adsorption process onto food-grade resin 33 .
To obtain a better insight into the adsorption process, the adsorption capacity was measured for more different initial concentrations of phenolic acids at three different temperatures (25, 37, and 45 °C, adjusted pH at 5.5), and three different pH values (pH 1.5, 5.5 and 10, adjusted temperature 37 °C). Those results are shown in c e vs q e diagrams ( Fig. 2  and 3). At different temperatures, all three phe nolic acids showed fluctuations with temperature (Fig. 2). All phenolic acids showed slightly higher adsorption capacity at 25 °C. With increasing temperature, the adsorption capacity decreased 32 . At different pH values, p-coumaric acid and chlorogenic acid showed somewhat higher adsorption capacity at pH 10 ( Fig. 3). In addition, it was found in literature that adsorption capacity increased with increasing pH 14 .
With the increasing concentration of phenolic acids, the adsorption capacity increased, but slight decrease in adsorption capacity was observed for concentration of 150 mg L -1 (Fig. 2 and 3). This could be connected with desorption that occurs in adsorption process 34 . At different temperatures, this could be seen for p-coumaric acid at 25 °C, caffeic acid at 25 and 37 °C, and chlorogenic acid at all temperatures (Fig. 2). At different pH values, this could be seen at pH 10 for p-coumaric acid, and at pH 1.5 and 10 for caffeic acid (Fig. 3).
The adsorption of p-coumaric acid, caffeic acid, and chlorogenic acid onto b-glucan showed some fluctuations with temperature. Adsorption capacity fluctuations with temperature can be connected to bond formation. It is known that hydrophobic interactions are an endothermic process 32,35 and accordingly, these interactions favor the increase in temperature. If the increase in temperature causes an increase in adsorption, it may indicate that mostly hydrophobic interactions are involved in the adsorption. H bond creation, on the other hand, is an exothermic process. Therefore, if the rising temperature causes decreasing adsorption, this may indicate the creation of H bonds 32,35 . Usually, both types of bonds exist in the interaction. The creation of different bonds could have affected the fluctuations of adsorption capacity with the temperature in our study. The creation of non-covalent bonds and interactions (like H bonds, Van der Waals forces and hydrophobic interactions) between phenolic acids and b-glucan in the temperature range from 25 to 45 °C, are in accordance with the predicted physical adsorption by adsorption isotherm models.
Furthermore, the influence of pH can be connected to chemical structure of phenolic acids, their functional groups, and the ratio of their deprotonated/protonated forms. Phenolic acids possess COOH groups as well as different number of OH groups. At lower pH, COOH and OH groups are in protonated (nonionic) forms. Protonated OH groups are important for the creation of H bonds at pH 1.5. As pH increases, the ratio of deprotonated/protonated functional groups (A -/A) increases as well. This means that the attraction forces between dissociated forms of phenolic acids and β-glucan at higher pH can be stronger, which can explain the higher adsorption capacity at higher pH in some cases (for p-coumaric acid and chlorogenic acid). The presence of phenolic acids in different protonated or deprotonated forms could have affected the adsorption at different pH 14,17,36 .

Adsorption isotherm models at different temperatures and pH values
The experimental results (c e vs q e ) were analyzed with five different nonlinear adsorption isotherm equations: Langmuir, Freundlich, Dubinin-Radushkevich, Temkin and Hill. The goal was to fit experimental data well with those equations, to determine their parameters with least possible error, and to possibly describe the adsorption process with those parameters. Namely, from all of these five models, parameters can be obtained: K L and q m from Langmuir, K F and 1/n from Freundlich, q s and E from Dubinin-Radushkevich, b T and A from Temkin, and q m , n H and K D from Hill. These parameters can predict useful, theoretical information about adsorption, such as the apparent maximum predicted adsorption capacity/saturation of β-glucan (q m and q s ), the apparent mean free energy (E) or whether there is a cooperativity in the adsorption between phenolic acids and β-glucan (n H ) 23 .  Table 2 reports parameters of isotherm models at three temperatures (25,37, 45 °C, pH 5.5). The parameters q m from both Langmuir and Hill and q s from Dubinin-Radushkevich can be used to gain insight into the model predicted apparent maximum adsorption capacities/saturation capacities of β-glucan 37,38 . The values of these parameters were from 518 to 778 mg g -1 at all temperatures, which is in the range of experimentally determined q e values. The parameter 1/n presents the intensity of the adsorption. For all phenolic acids, the 1/n was lower than 1, which suggests that the adsorption was favored for all phenolic acids at all three temperatures. The parameter E represents apparent mean   free energy of adsorption. All free energies of adsorption were lower than 8000 J mol -1 , which indicates a physical bond creation between phenolic acids and β-glucan, like H bonds or Van der Waals forces 23 at all three temperatures. The creation of these non-covalent bonds is in agreement with earlier studies 14,18,32 . The parameter n H is lower than 1 for all studied phenolic acids at all temperatures, which suggests a negative cooperation in the adsorption. This means that when a molecule of phenolic acid is adsorbed onto the surface of β-glucan, the affinity for other molecules decreases. Temkins parameters b T and A are shown in relation to q e , in Fig. 3. The parameter b T , Temkin's apparent heat of adsorption, and parameter A, apparent maximum bonding energy, both decreased as q e increased, suggesting that compounds with higher maximum adsorption capacities had lower apparent adsorption heat (Fig. 3). Table 3 reports parameters of isotherm models at three pH values (pH 1.5, 5.5, 10, at 37 °C). Parameters q m from Langmuir and Hill, and q s from Dubinin-Radushkevich isotherm were in the range from 460 to 1545 mg g -1 , which is in the range of experimentally determined q e values or somewhat higher. For all phenolic acids, the 1/n was lower than 1, which suggested that the adsorption was favored for all phenolic acids at all pH. The mean free energy of adsorption E was lower than 8000 J mol -1 at all pH, which still indicated physical bonds between phenolic acids and β-glucan. The apparent heat of the adsorption b T and bonding energy A, decreased as q e increased, suggesting that compounds with higher maximum adsorption capacities had lower apparent adsorption heat (Fig. 4). The n H was lower than 1 at all pH values, which suggested a negative cooperation in adsorption. As already mentioned, this means that when a molecule of polyphenol is adsorbed onto the surface of β-glucan, the affinity for other molecules decreases.
Furthermore, predicted parameters q m from Langmuir and Hill models, q s from Dubinin-Radushkevich model were correlated to experimentally determined q e to see how well the predicted values described the experimental values (Fig. 5). Predicted values q m and q s correlated well with the experi-mentally determined q e at different temperatures and pH, with high R 2 (0.9562 to 0.9999). This suggested well conducted modelling process. Thermodynamics Thermodynamic parameter ΔG 0 is shown in Table 4, together with ΔS 0 and ΔH 0 obtained from the Van't Hoff plots. It should be mentioned that the calculation of these thermodynamic parameters involves the use of constants from Langmuir isotherm (K L (L mol -1 )), which were determined with modelling, and depend on precisely conducted modelling process. This could affect the determination of thermodynamic parameters. Since K L values showed differences with temperature (Table 2), they were used in this study for the evaluation of thermodynamic parameters. For all phenolic acids, the standard reaction Gibbs energy ΔG 0 was negative at all temperatures, which was similar to literature data 23,39 and suggested a spontaneous process. Van't Hoff plots were then used to evaluate ΔS 0 and ΔH 0 . More precisely, standard reaction enthalpy ΔH 0 was calculated from the slope, and standard reaction entropy ΔS 0 from the intercept of the Van't Hoff plot. Lines in Van't Hoff plots had R 2 from low values (0.0136) to higher values (0.7202) ( Table 4), which suggested that the lines were not linear in all cases (p-coumaric acid), and that they should be interpreted with caution. This could be connected to the complexity of adsorption process 38 . Nevertheless, ΔS 0 and ΔH 0 were obtained from slopes and intercepts of Van't Hoff plots. The results showed the positive value of standard reaction entropy ΔS 0 , which supported a spontaneous process in the interaction of phenolic acids with β-glucan (except for chlorogenic acid). ΔH 0 was found to be negative, indicative of an exothermic reaction 35 .

Conclusions
By studying adsorption, this study showed that phenolic acids interacted with β-glucan in a complex process affected by the concentration of phenolic acids, chemical structure of phenolic acids or pH value and temperature of environment. The Ta b l e 4 -Thermodynamic parameters for the adsorption of phenolic acids and β-glucan Polyphenol ΔG 0 (J mol -1 ) R 2 ΔH 0 (J mol -1 ) ΔS 0 (J mol -1 K -1 ) 25  *in these cases, the data were inconsistent with a linear fit analysis of adsorption of selected phenolic acids onto β-glucan indicated that temperature and pH affected the adsorption process causing some fluctuations in adsorption capacities. Furthermore, the adsorption capacity was higher for p-coumaric acid and chlorogenic acid at pH 10. Experimental results in the adsorption process can be modelled well with adsorption isotherms, which can give useful information about the adsorption process. According to adsorption isotherm parameters, the adsorption process could be physical and noncooperative at all temperatures and pH values. It can be suggested that bonds created between all phenolic acids and β-glucan were physical bonds like H bonds and Van der Waals forces. According to thermodynamic parameters, the adsorption process was spontaneous (except for chlorogenic acid) and exothermic for all phenolic acids. All this might have an important effect on the polyphenol bioactivities in the human organism. Further studies are needed to completely understand interactions.