Identification and Capture of Phenolic Compounds from a Rapeseed Meal Protein Isolate Production Process By-Product by Macroporous Resin and Valorization Their Antioxidant Properties

In this study, phenolic compounds from an aqueous protein by-product from rapeseed meal (RSM) were identified by HPLC-DAD and HPLC-ESI-MS, including sinapine, sinapic acid, sinapoyl glucose, and 1,2-di-sinapoyl gentibiose. The main phenolic compound in this by-product was sinapine. We also performed acid hydrolysis to convert sinapine, and sinapic acid derivatives present in the permeate, to sinapic acid. The adsorption of phenolic compounds was investigated using five macroporous resins, including XAD4, XAD7, XAD16, XAD1180, and HP20. Among them, XAD16 showed the highest total phenolic contents adsorption capacities. The adsorption behavior of phenolic compounds was described by pseudo-second-order and Langmuir models. Moreover, thermodynamics tests demonstrated that the adsorption process of phenolic compounds was exothermic and spontaneous. The highest desorption ratio was obtained with 30% (v/v) and 70% (v/v) ethanol for sinapine and sinapic acid, respectively, with a desorption ratio of 63.19 ± 0.03% and 94.68 ± 0.013%. DPPH and ABTS tests revealed that the antioxidant activity of the hydrolyzed fraction was higher than the non-hydrolyzed fraction and higher than the one of vitamin C. Antioxidant tests demonstrated that these phenolic compounds could be used as natural antioxidants, which can be applied in the food industry.


Introduction
Rapeseed (Brassica napus L.) is an oilseed cultivated worldwide, particularly in Europe and China (11.684 and 11.760 tons/year, respectively) [1]. This resource shows phenolic compound content as high as 1.705 g of sinapic acid equivalent (SAE)/100 g of dry matter [2]. Naczk et al. found that the proportion of rapeseed polar phenolic compounds was ten times higher than other oilseeds [3]. Rapeseed phenolic compounds are mainly composed of sinapic acid esters. Sinapic acid is a hydroxycinnamic acid widespread in plants [4]. The most abundant rapeseed phenolic compound is an ester of sinapic acid and choline called sinapine [5,6]. Other hydrophilic non-phenolic parts may be associated with sinapic acids such as glucose-based osides (such as sinapoyl glucose) associated or not with kaempferol (such as kaempferol-sinapoyl-trihexoside) [7][8][9]. Rapeseed also contains minor phenolic compounds such as free sinapic acid, ferulic acid, vanillic acid, or syringic acid [3,10]. Figure 1A shows the SE-HPLC chromatogram at 325 nm of the ultrafiltration (UF) permeate. This permeate was obtained from a rapeseed albumin isolate production process recently patented [27,28]. The main peak can be observed at 18.94 min of retention time alongside marginal signals at 23.31, 28.33, and 31.68 min. ESI-MS analysis revealed that this peak contained molecules with m/z of 310 (positive mode), 385, 753, and 223 (negative mode) (Table 1). This corresponded to sinapine (SP), sinapoyl glucose (SG), 1,2-di-sinapoyl gentiobise, and sinapic acid (SA), respectively. The chemical structures of these phenolic compounds are presented in Figure 2. The identification of sinapine and sinapic acid was confirmed by injection of standards. R. Khattab et al. [8] also identified SP, SA, and SG from canola/rapeseed organic solvent extracts such as 70% ethanol, 70% methanol, or 70% iso-propanol (v/v). In the UF permeate, SP, other identified SA esters and SA represented 59.78%, 21.66%, and 2.84% in proportion, respectively. Only 15.72% of the phenolic compounds remained unidentified. These observations also agreed with the report from U. Thiyam-Holländer et al. [16]. In their study, the de-oil rapeseed was extracted by distilled water at a ratio of 1:10 (w/v) for 10 min. The unidentified compounds were most probably other polar sinapic acid esters.
Molecules 2021, 26, x FOR PEER REVIEW 3 of 21 considered to choose the most appropriate eluent for phenolic compounds captured from this effluent. Finally, the antioxidant activity of the phenolic fractions obtained after desorption was assessed and compared to sinapine, sinapic acid, and vitamin C standards. Figure 1A shows the SE-HPLC chromatogram at 325 nm of the ultrafiltration (UF) permeate. This permeate was obtained from a rapeseed albumin isolate production process recently patented [27,28]. The main peak can be observed at 18.94 min of retention time alongside marginal signals at 23.31, 28.33, and 31.68 min. ESI-MS analysis revealed that this peak contained molecules with m/z of 310 (positive mode), 385, 753, and 223 (negative mode) (Table 1). This corresponded to sinapine (SP), sinapoyl glucose (SG), 1,2-disinapoyl gentiobise, and sinapic acid (SA), respectively. The chemical structures of these phenolic compounds are presented in Figure 2. The identification of sinapine and sinapic acid was confirmed by injection of standards. R. Khattab et al. [8] also identified SP, SA, and SG from canola/rapeseed organic solvent extracts such as 70% ethanol, 70% methanol, or 70% iso-propanol (v/v). In the UF permeate, SP, other identified SA esters and SA represented 59.78%, 21.66%, and 2.84% in proportion, respectively. Only 15.72% of the phenolic compounds remained unidentified. These observations also agreed with the report from U. Thiyam-Holländer et al. [16]. In their study, the de-oil rapeseed was extracted by distilled water at a ratio of 1:10 (w/v) for 10 min. The unidentified compounds were most probably other polar sinapic acid esters.      Figure 1B shows SP and SA kinetics during acidic hydrolysis of the UF permeate. Initial SP concentration was around 0.8 mM (i.e., 0.25 g/L) and continuously decreased during 180 min to reach less than 0.1 mM. SA concentration increased dramatically during the first 60 min and stabilized after 90 min at around 0.35 mM. During this period, the amount of SA increased from 23.35 mg/100 g to 135.18 mg/100 g rapeseed meal. This suggests that the chemical hydrolysis of SP ester bond into SA reached an equilibrium. Hence, the SP concentration decrease after 90 min was surprising. This indicated a degradation into other molecules with no UV properties at 325 nm-this can be hypothesized as a degradation of the benzene ring. Interestingly, at 180 min the purity of SA was the highest (63.66%). SP was still present but at a low proportion (17.44%). Subsequently, this reaction duration was chosen for further study of phenolic compounds capture and bioactivity characterization.

Identification of Phenolic Compounds in the Ultrafiltration (UF) Permeate
A. Siger et al. [10] found that the contents of sinapic acid after acidic hydrolysis of two rapeseed extracts from two cultivars were 93.2 mg/100 g (cv. Visby) and 89.4 mg/100 g (cv. Bellevue). The SA content observed in our study is slightly higher (135.18 mg/100 g). This could be due to a difference in starting material or to the solvent used for extraction. Indeed, these authors used methanol while we used an aqueous solvent (NaCl 0.11 M at pH 2). This could also be due to a difference in the hydrolysis step duration (during 20 min at 90 • C).

Method Validation
The validation of the SE-HPLC method was performed according to the International Conference on Harmonisation (ICH) of technical requirements for registration of pharmaceuticals for human use [29] recommendations. All experiments were conducted in triplicate. The parameters such as LOD (limit of detection), LOQ (limit of quantification), and standard deviation (SD) values were determined and are summarized in Table 2 using pure standards such as sinapine and sinapic acid. As listed in Table 2, the LODs of sinapine and sinapic acid were 0.25 µg/mL. The LOQs of sinapine and sinapic acid were 0.6 µg/mL. These low values of LOD and LOQ demonstrated a high sensitivity of the developed SE-HPLC technical method. The linearity of calibration curves was constructed with three measurements for each calibration point ranging from 0.05 to 1.25 mg/mL. A strong linear correlation (R 2 > 0.99) between the concentration and the peak area was observed. Preliminary results showed that the retention time and peak area were almost similar with intra-and inter-day standard deviation (SD) values less than 0.12 and 1.9 for the retention time and peak area, respectively. These results revealed the good performance of the developed analytical method for quantification of phenolic compounds such as sinapine and sinapic acid within the permeate with and without hydrolysis. Figure 3 shows the adsorption capacity of total phenolic content (TPC), sinapine, and sinapic acid from permeate and hydrolyzed permeate on XAD4, XAD7, XAD16, XAD1180, and HP20 resins. Table 7 displays the properties of the resins. XAD4, XAD16, XAD1180, and HP20 are nonpolar resins made out of SDVB polymers. They differ by bead and pore size (50-300 Å). XAD7 is composed of acrylamide polymer which is considered a mild polar [26]. All these resins were demonstrated to adsorb phenolic compounds from many different plant extracts [30,31]. The adsorption of TPC from non-hydrolyzed permeate ranged from about 11 (for XAD4 and XAD7) to 12.5 mgSAE/g (for XAD16) of dry resins ( Figure 3A). Interestingly, the adsorption of sinapine (main phenolic compound in nonhydrolyzed permeate) was similar to TPC, ranging from 11 (for XAD7) to 13.5 mg/g (for XAD16) of dry resin ( Figure 3A). Molecules 2021, 26, x FOR PEER REVIEW 6 of 21 Notably, XAD4 showed the lowest adsorption capacity of TPC (about 3.7 mg SAE/g of dry resin) and of sinapic acid (about 3.25 mg/g of dry resin) in hydrolyzed permeate ( Figure 3B). On the other hand, XAD16 showed the highest adsorption capacity of TPC (4.1 mgSAE/g of dry resin) and of sinapic acid (3.9 mg/g of dry resin) ( Figure 3B).

Resin Screening
The adsorption capacity depends on a molecule's affinity toward the material and the specific surface [32]. This last parameter is conditioned by bead size and can be modulated by pore size if it limits the molecule's inner accessibility [26]. Pore ≤ 50 Å was shown to limit the diffusion of phenolic compounds into particle pores by steric hindrance [29,32]. Among styrene-divinyl benzene (SDVB) resins, XAD16 showed the highest specific area (900 m²/g) followed by XAD4, XAD1180, and HP20. Interestingly, this ranking corresponded to observed capacity values. This indicates that SDVB resin capacities were essentially governed by the specific area. It can be hypothesized that XAD4 low pore size (50 Å) did not yield additional diffusion limitations. The adsorption behavior of phenolic compounds in non-hydrolyzed and hydrolyzed permeates onto SDVB resins and mild polar resin (XAD7) had the same pattern. Probably, the interaction of sinapine ( Figure 2) (a main phenolic compound in non-hydrolyzed permeate) and sinapic acid (a main phenolic compound in hydrolyzed permeate) ( Figure 2) and SDVB resins were through pi-pi stacking interaction (aromatic ring part of phenolic compounds and benzene rings of SDVB resins) [26]. XAD16 and XAD1180 showed the highest adsorption capacity of TPC from non-hydrolyzed and hydrolyzed permeates. However, XAD16 had the highest adsorption capacity of sinapine and sinapic acid. The very high specific area of XAD16 (900 m²/g) showed better adsorption in terms of massic capacity than XAD7 or HP20 with a similar specific area (450 or 500 m²/g). So, the adsorption kinetics, isotherms, thermodynamics properties, and phenolic compounds desorption were further investigated with this resin.

Adsorption Kinetics
Adsorption kinetics of the main phenolic compounds from UF permeate with (sinapic acid) or without hydrolysis (sinapine) on XAD16 are presented in Figure 4A-B. Very similar trends were observed. A large part of phenolic compounds was quickly adsorbed (up to 90% of the equilibrium capacity in 30 min). Then, the kinetics slowed down up to the equilibrium observed between 60 min and 120 min. As expected from the resin screening results, this indicated that XAD16 has a strong affinity toward rapeseed phenolic compounds. Such behavior was also reported for adlay bran phenolic compounds [30]. Notably, XAD4 showed the lowest adsorption capacity of TPC (about 3.7 mg SAE/g of dry resin) and of sinapic acid (about 3.25 mg/g of dry resin) in hydrolyzed permeate ( Figure 3B). On the other hand, XAD16 showed the highest adsorption capacity of TPC (4.1 mgSAE/g of dry resin) and of sinapic acid (3.9 mg/g of dry resin) ( Figure 3B).
The adsorption capacity depends on a molecule's affinity toward the material and the specific surface [32]. This last parameter is conditioned by bead size and can be modulated by pore size if it limits the molecule's inner accessibility [26]. Pore ≤ 50 Å was shown to limit the diffusion of phenolic compounds into particle pores by steric hindrance [29,32]. Among styrene-divinyl benzene (SDVB) resins, XAD16 showed the highest specific area (900 m 2 /g) followed by XAD4, XAD1180, and HP20. Interestingly, this ranking corresponded to observed capacity values. This indicates that SDVB resin capacities were essentially governed by the specific area. It can be hypothesized that XAD4 low pore size (50 Å) did not yield additional diffusion limitations. The adsorption behavior of phenolic compounds in non-hydrolyzed and hydrolyzed permeates onto SDVB resins and mild polar resin (XAD7) had the same pattern. Probably, the interaction of sinapine ( Figure 2) (a main phenolic compound in non-hydrolyzed permeate) and sinapic acid (a main phenolic compound in hydrolyzed permeate) ( Figure 2) and SDVB resins were through pi-pi stacking interaction (aromatic ring part of phenolic compounds and benzene rings of SDVB resins) [26]. XAD16 and XAD1180 showed the highest adsorption capacity of TPC from non-hydrolyzed and hydrolyzed permeates. However, XAD16 had the highest adsorption capacity of sinapine and sinapic acid. The very high specific area of XAD16 (900 m 2 /g) showed better adsorption in terms of massic capacity than XAD7 or HP20 with a similar specific area (450 or 500 m 2 /g). So, the adsorption kinetics, isotherms, thermodynamics properties, and phenolic compounds desorption were further investigated with this resin.

Adsorption Kinetics
Adsorption kinetics of the main phenolic compounds from UF permeate with (sinapic acid) or without hydrolysis (sinapine) on XAD16 are presented in Figure 4A,B. Very similar trends were observed. A large part of phenolic compounds was quickly adsorbed (up to 90% of the equilibrium capacity in 30 min). Then, the kinetics slowed down up to the equilibrium observed between 60 min and 120 min. As expected from the resin screening results, this indicated that XAD16 has a strong affinity toward rapeseed phenolic compounds. Such behavior was also reported for adlay bran phenolic compounds [30].  Adsorption kinetics were regressed with pseudo-first-order (PFO, [33]) and pseudosecond-order (PSO, [34]) equations in linearized form ( Figure 4C-F). Table 3 summarizes the corresponding equations, model parameter values, and R 2 of the linear regressions. Table 3. Kinetic parameters for sinapine and sinapic acid adsorption onto XAD16 resin in non-hydrolyzed and hydrolyzed permeates.

Kinetic Model Equation Parameter
Phenolic Compound

Sinapine Sinapic Acid
Pseudo-first-order 2587 q e,exp (mg/g) 13.532 3.861 q e,cal (mg/g) 13.544 3.861 R 2 1 1 The R 2 obtained with linearized PFO (ln(q e −q t ) vs. t) was less than 0.94 for both sinapine and sinapic acid. Furthermore, the calculated q e for XAD16 (0.32 mg/g and 0.29 mg/g for sinapine and sinapic acid, respectively) was found to be very different from experimental values (13.52 and 3.86 mg/g for sinapine and sinapic acid, respectively). On the other hand, the R 2 of the linear regression of the t/q t vs. t plot was very close or equal to 1 (0.999 and 1). Moreover, the calculated q e (13.54 and 3.86 mg/g for sinapine and sinapic acid) predicted from the PSO model was very near the experimental values of q e (13.53 and 3.86 mg/g for sinapine and sinapic acid). These results indicate that adsorption kinetics followed a PSO model for the XAD16 resin. This was also observed with adsorption of chlorogenic acid from by-product protein isolates on XAD16 resin [26].
Solute transport phenomena are complex in adsorption processes. In the liquid phase, solutes are transported by convection and diffusion. There is also diffusive transport from the liquid phase to the bead surface (through a limit liquid film) and diffusive transport inside the particle's pores. Adsorption kinetics may be modulated by several diffusional types of transports. The intra-particle diffusion model [35] is commonly used to investigate the diffusive rate-controlling phenomenon [25,36]. Figure 4G,H show q t vs. t 0.5 plots obtained with the XAD16 resin. These plots correspond to the linear form of the intra-particle diffusion model ( Table 2). The slopes represent the constant rate (k i ) of each adsorption step while C i (intercept at y-axis) is related to the thickness of the limiting layer. R 2 , k i , and C i values obtained from linear regressions are displayed in Table 3. Interestingly, k 1 (0.103 and 0.09 (mg/g)/min 0.5 ) for sinapine and sinapic acid, respectively) are by far higher than k 2 (approximately 0). These differences might come from the smaller molecular size of sinapic acid than sinapine. It can also be noticed that R 2 values for sinapine and sinapic acid are 0.9828 and 0.9639, respectively. This indicates that for both liquid effluents, the adsorption process is limited by two diffusional effects. In the previous study, a similar pattern was also observed with the adsorption of chlorogenic acid onto XAD16 resin [26]. Very similar results were observed with the adsorption of alfalfa phenolic compounds on HP20 and AER1 resins [25]. It was interpreted as a two steps adsorption process. The first one is related to the diffusional transport throughout the boundary layer at the liquid/bead interface. Its high rate constant (K i,1 was 1.03 and 0.09 (mg/g)/min 0.5 ) for sinapine and sinapic acid, respectively) indicates a low diffusional limitation at this stage. The second one is due to intraparticle diffusion.
The low rate constant (K i,2 approximately equal to 0 for both liquid effluents) indicates a stronger diffusional limitation. Such observations and explanations were also made by others [31,37].
In this study, our results revealed that the adsorption capacity could reach the equilibrium state after 30 min of adsorption time. The adsorption of phenolic compounds such as sinapine and sinapic acid was followed by a pseudo-second-order model. As discussed in our previous study [26], the adsorption of these phenolic compounds is based on the pi-pi stacking interaction between the benzene ring of phenolic compounds and benzene rings macroporous resin (XAD16). Therefore, this adsorption was considered as spontaneous and physical adsorption [26]. Figure 5 shows adsorption isotherms of sinapine and sinapic acid onto XAD16 at 25 • C. Data were regressed with Langmuir ( Figure 5A,C) and Freundlich ( Figure 5B,D) equations as previously carried out elsewhere [24,38,39]. Table 4 lists the R 2 of the regression, equations, and model parameters with sinapine and sinapic acid. R 2 values indicated that experimental data were better fitted by the Langmuir model (0.997 and 0.9999 for sinapine and sinapic acid, respectively) than by the Freundlich model (0.996 and 0.9978 for sinapine and sinapic acid, respectively). This indicated that the same adsorption mechanism took place in any case. This consisted of monolayer adsorption of phenolic compounds at the surface of the resin [24,38,39]. These findings are also in agreement with another study on chlorogenic acid adsorption on XAD16 resin from sunflower meal [26].  Moreover, this finding is consistent with that of M. Moreno-González et al. [21] regarding the adsorption of sinapic acid from canola/rapeseed meal using the FPX66 resin. The maximum adsorption capacity of sinapine and sinapic acid was 35.93 mg/g (0.12 mol/g) and 23.96 mg/g (0.11 mol/g) of dry resin, respectively ( Table 4). The values obtained in this study suggest that the adsorption onto resin material implies interaction with the same part of the molecule (sinapine and sinapic acid) without the interference of the choline part (quaternary ammonium group, Figure 2). This observation has never been published. Meanwhile, according to M. Moreno-González et al. [21], the maximum adsorption capacity for sinapic acid was lower (about 15 mg/g or 0.07 mol/g) onto the nonpolar (SDVB) FPX66 resin. The different values might be due to other organic compounds such as carbohydrates, amino acids, and proteins in raw materials. It might also be due to the impact of ionic strength. Moreover, the difference in physical characteristics also caused the difference in adsorption capacity. A. Thiel et al. [40] investigated the adsorption of sinapic acid and other compounds onto zeolites and hydrophobic resins, including XAD16, the resin used in this study. These authors claimed that the adsorption capacity of sinapic acid was higher than the one presented here (44.5 mg/g or 0.20 mol/g of dry resin). However, these authors evaluated the experiments at different pH (pH 5) and used different starting materials; rapeseed meal extracted with heated deionized water at 70 • C at a ratio of 1:8 (solid:liquid) that might explain the difference in these values.

Determinations of Thermodynamic Parameters
The effect of temperature on the adsorption capacity of phenolic compounds from the XAD16 resin in the two liquid effluents was investigated to obtain the thermodynamic parameters of adsorption. Langmuir model parameters and R 2 are listed in Table 5; ∆H and ∆S were determined through the slope and intercept of ln K L against 1/T (Equations (2) and (3)) ( Figure 6A,B) according to Van Hoff's equation. Enthalpy changes (∆H) for the sinapine and sinapic acid adsorption process on XAD16 resin were −2.56, and −2.72 kJ/mol, respectively (Table 5). Negative values indicate an exothermic adsorption process. The fact that values were less than 43 kJ/mol demonstrates that the adsorption process of phenolic compounds on the XAD16 resin was governed by physical rather than chemical interactions [30]. This demonstrates that the XAD16 resin would not undergo structural changes during the phenolic compounds' adsorption process. Therefore, the adsorption of phenolic compounds on the resin only takes place through a physical mechanism with no chemical reactions. This observation was also reported by Z. P. Gao et al. [41] who studied the adsorption of polyphenols from kiwifruit juice using AB-8 resin [41]. In addition, the entropy changes (∆S) values of XAD16 were −55.95 and −8.37 kJ/molK for sinapine and sinapic acid, respectively. These negative values suggest a random adsorption process at the solid-liquid interface [41] which occurred due to the desorption process of water molecules previously adsorbed onto the resins' surface [41]. The negative free energy change (∆G) deduced from ∆H and ∆S (Table 5) suggests that phenolic compounds' adsorption onto the XAD16 resin was a spontaneous process. Moreover, the absolute value of ∆G < 20 kJ/mol confirmed the physical adsorption of phenolic compounds onto XAD16 resin [41,42]. adsorption onto the XAD16 resin was a spontaneous process. Moreover, the absolute value of ∆G < 20 kJ/mol confirmed the physical adsorption of phenolic compounds onto XAD16 resin [41,42].

Desorption of Phenolic Compounds from the XAD16 Resin
To assess the effect of ethanol concentration on desorption, five ethanol concentrations were evaluated: 30, 50, 70, and 90% (v/v). Preliminary results showed that the desorption curve reaches equilibrium after 120 min (data not shown). As shown in Figure 7, ethanol concentration significantly influenced SP and SA desorption ratio. The highest desorption ratio of sinapine and sinapic acid was observed with ethanol (30% and 70% (v/v) (p < 0.05), respectively). These results indicate that the desorption ratio was influenced by the ethanol concentration and the solubility of phenolic compounds in the desorption phase. Indeed, sinapine is a polar compound demonstrating a good desorption ratio with ethanol at a lower concentration. Sinapic acid is by far less polar, therefore, presenting a higher desorption ratio at a higher ethanol concentration. Surprisingly, the desorption ratios of sinapic acid and sinapine in this study were much higher than in the report of A. Thiel et al. [14,40] (about 44% and 3.7% for SA and SP, respectively) who used the same macroporous resin (XAD16) and desorption with 70% aqueous ethanol. However, a high desorption ratio of sinapic acid was also found by M. Moreno-González et al. [21] with 70% (v/v) of ethanol.

Desorption of Phenolic Compounds from the XAD16 Resin
To assess the effect of ethanol concentration on desorption, five ethanol concentrations were evaluated: 30, 50, 70, and 90% (v/v). Preliminary results showed that the desorption curve reaches equilibrium after 120 min (data not shown). As shown in Figure 7, ethanol concentration significantly influenced SP and SA desorption ratio. The highest desorption ratio of sinapine and sinapic acid was observed with ethanol (30% and 70% (v/v) (p < 0.05), respectively). These results indicate that the desorption ratio was influenced by the ethanol concentration and the solubility of phenolic compounds in the desorption phase. Indeed, sinapine is a polar compound demonstrating a good desorption ratio with ethanol at a lower concentration. Sinapic acid is by far less polar, therefore, presenting a higher desorption ratio at a higher ethanol concentration. Surprisingly, the desorption ratios of sinapic acid and sinapine in this study were much higher than in the report of A. Thiel et al. [14,40] (about 44% and 3.7% for SA and SP, respectively) who used the same macroporous resin (XAD16) and desorption with 70% aqueous ethanol. However, a high desorption ratio of sinapic acid was also found by M. Moreno-González et al. [21] with 70%  Figure 8 shows the HPLC chromatogram of phenolic compounds after purification with the XAD16 resin. The peaks of sinapine ( Figure 8A) and sinapic acid ( Figure 8B) are highlighted in the desorption fractions. Sinapine (63.31%) and sinapic acid (73.00%) are the mains phenolic compounds in the hydrolyzed fraction after purification. As different compositions of phenolic compounds after purification might lead to different biological actions, we used both fractions to assess the antioxidant activity in subsequent experiments.

In Vitro Antioxidant Activity
The free radical scavenging rate of DPPH was determined to assess the antioxidant activity of the non-hydrolyzed and hydrolyzed fractions by comparison to pure SP, SA, and vitamin C. As shown in Figure 9A, the scavenging rate increased with the concentration of non-hydrolyzed fraction (N fraction). The same was observed for the hydrolyzed fraction (H fraction), but in this case, interestingly, we observed scavenging rates higher than the ones obtained with vitamin C at 5, 10, and 20 µg/mL. These results were confirmed by the ABTS assay ( Figure 9B).   Figure 8 shows the HPLC chromatogram of phenolic compounds after purification with the XAD16 resin. The peaks of sinapine ( Figure 8A) and sinapic acid ( Figure 8B) are highlighted in the desorption fractions. Sinapine (63.31%) and sinapic acid (73.00%) are the mains phenolic compounds in the hydrolyzed fraction after purification. As different compositions of phenolic compounds after purification might lead to different biological actions, we used both fractions to assess the antioxidant activity in subsequent experiments.

In Vitro Antioxidant Activity
The free radical scavenging rate of DPPH was determined to assess the antioxidant activity of the non-hydrolyzed and hydrolyzed fractions by comparison to pure SP, SA, and vitamin C. As shown in Figure 9A, the scavenging rate increased with the concentration of non-hydrolyzed fraction (N fraction). The same was observed for the hydrolyzed fraction (H fraction), but in this case, interestingly, we observed scavenging rates higher than the ones obtained with vitamin C at 5, 10, and 20 µg/mL. These results were confirmed by the ABTS assay ( Figure 9B).

In Vitro Antioxidant Activity
The free radical scavenging rate of DPPH was determined to assess the antioxidant activity of the non-hydrolyzed and hydrolyzed fractions by comparison to pure SP, SA, and vitamin C. As shown in Figure 9A, the scavenging rate increased with the concentration of non-hydrolyzed fraction (N fraction). The same was observed for the hydrolyzed fraction (H fraction), but in this case, interestingly, we observed scavenging rates higher than the ones obtained with vitamin C at 5, 10, and 20 µg/mL. These results were confirmed by the ABTS assay ( Figure 9B).  Table 6 presents IC50 values of compounds tested using DPPH and ABTS assays. These data demonstrate that the antioxidant activity of the hydrolyzed phenolic fraction was stronger than the one of the non-hydrolyzed fraction and vitamin C in both assays. These findings are likely due to the enrichment of the hydrolyzed fraction in sinapic acid as this compound exhibited the lowest IC50 values. It has already been shown that phenolic compounds found in RSM are excellent antioxidants. Previous studies showed that the antioxidant activity is most likely due to vinylsyringol found in rapeseed oil [16,43]. In this study, we show that sinapic acid also has strong antioxidant properties. Interestingly, the hydrolyzed fraction (containing mainly SA) was more effective than the non-hydrolyzed fraction (mainly containing SP) and vitamin C. These observations are in accordance with the report of S. Vourela et al. [44] who investigated the antioxidant function of phenolic compounds in rapeseed oil. Other researchers [10,12,45] and our previous study [46] also indicated that the high concentration of phenolic compounds is related to a high antioxidant capacity. Taken together, these data indicate that phenolic compounds isolated from RSM are interesting natural antioxidants that could be used in the food industry or other applications [47].

Materials
The rapeseed meal was provided by Olead (Pessac, Bordeaux, France). Sinapine standard was purchased from ChemScience (JJ08852, Monmouth Junction, NJ, USA). Sinapic acid and formic acid (FA) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Sodium chloride (NaCl) and sodium hydroxide (NaOH) pellets were purchased from VWR (Radnor, PA, USA). Chlorohydric acid (HCl) 37% was from CarloErba (Val-de-Reuil, France). Acetonitrile (ACN) solution was provided by Biosolve BV (Valkensward, Figure 9. Scavenging activity of the non-hydrolyzed (N fraction) and hydrolyzed (H fraction) fractions compared to pure sinapine, sinapic acid, and vitamin C determined using (A) DPPH assay and (B) ABTS assay. Bars labeled with the different lowercase letters and uppercase letters are significantly different (p < 0.05). Table 6 presents IC50 values of compounds tested using DPPH and ABTS assays. These data demonstrate that the antioxidant activity of the hydrolyzed phenolic fraction was stronger than the one of the non-hydrolyzed fraction and vitamin C in both assays. These findings are likely due to the enrichment of the hydrolyzed fraction in sinapic acid as this compound exhibited the lowest IC50 values. Table 6. IC50 values of compounds tested using the DPPH and ABTS assays. It has already been shown that phenolic compounds found in RSM are excellent antioxidants. Previous studies showed that the antioxidant activity is most likely due to vinylsyringol found in rapeseed oil [16,43]. In this study, we show that sinapic acid also has strong antioxidant properties. Interestingly, the hydrolyzed fraction (containing mainly SA) was more effective than the non-hydrolyzed fraction (mainly containing SP) and vitamin C. These observations are in accordance with the report of S. Vourela et al. [44] who investigated the antioxidant function of phenolic compounds in rapeseed oil. Other researchers [10,12,45] and our previous study [46] also indicated that the high concentration of phenolic compounds is related to a high antioxidant capacity. Taken together, these data indicate that phenolic compounds isolated from RSM are interesting natural antioxidants that could be used in the food industry or other applications [47].

Rapeseed Protein Isolate and Its Aqueous By-Product Production
The protein isolate/purification process was performed in two steps: First, protein extraction from rapeseed meal (RSM), and second, protein purification by ultrafiltration as recommended in [27,28]. For protein extraction, the appropriate amount of RSM and NaCl 0.1 M solution was mixed at a solid/liquid ratio of 1:9 (w/w). The extraction step was performed at pH 2 under agitation at 400 rpm for 30 min. Then, the liquid extract was clarified by centrifugation at 10,000 g for 30 min at 20 • C (Heraeus Megafuge 16R centrifugation system, ThermoScientific, Waltham, MA, USA). The supernatant was further clarified by tangential microfiltration using a 0.22 µm membrane at 0.2 bar of transmembrane pressure (0.1 m 2 area, Pellicon Mini Cassette Durapore, polyvinylidene difluoride (PVDF) membrane, MA, USA). The protein purification was achieved by diafiltration using an Äkta Flux 6 device (GE Healthcare, Chicago, IL, USA) at a feed flow of 2 L/min and a transmembrane pressure of 1.5 bars. During the process, the retentate compartment was fed with 0.5 M salt solution at the same flow rate as the permeate flux in order to keep the volume constant. Six diavolume (DV, DV = V NaCl /V o ) of NaCl solution was poured. Then, three DV of deionized water was used to flush NaCl from proteins. The permeate containing the phenolic compounds was recovered and adjusted to pH 2 by adding HCl 1M for further studies.

Phenolic Compounds Hydrolysis
Phenolic compounds contained in the UF permeate were hydrolyzed under acidic conditions as suggested by A. Siger et al. [10]. To do so, 350 mL of permeate was mixed with 70 mL of HCl 37% (v/v). The reaction was carried out at 75 • C with agitation at 150 rpm for 3 h. During the reaction, 600 µL of samples were taken at 30, 60, 90, 120, 150, and 180 min for phenolic compounds analysis and quantification by SE-HPLC.

Phenolic Compounds Identification by HPLC-MS
Phenolic compounds from UF permeate with or without acidic hydrolysis were separated by SE-HPLC and identified by ESI-MS according to T. T. Le et al. [26,46] SA and SP were quantified from UF permeate with or without acidic hydrolysis by the above described SE-HPLC method using their UV signal at 325 nm. To do so, calibration curves were established by quantifying the peak area of standard SA and SP at concentrations ranging from 0.05 to 1.25 mg/mL. The linear regression equations of the calibration curves were y = 2.9.10 7 x and y = 5.31.10 7 x with R 2 = 0.9998 and R 2 = 0.9984 of SP and SA, respectively.

Method Validation
Validation of this method was conducted according to the International Conference on Harmonization (ICH) [29]. Linearity of the detector responses and limits were performed in the range 0.05-1.25 mg/mL. Each calibration plot performs an average of three independent repetitions for five different concentration levels. Linear regression analysis was used to determine the slope and correlation coefficient (R 2 ). All experiments were tested three times for phenolic compounds standards and the phenolic compounds presenting in samples solution in this study.

Total Phenolic Contents (TPC)
The quantification of total phenolic content (TPC) was performed by the "sum of phenolic acids" method of R. Khattab et al. [8]. TPC was estimated as SA equivalent (SAE) from the sum area of all peaks of phenolic compounds at 325 nm using the above described SE-HPLC method.

NaCl Content
NaCl content in UF permeate was evaluated by measuring each sample's conductivity using a conduct meter (MeterLab HPM210, Radiometer analytical, Lyon, France). NaCl solutions with concentrations ranging from 0.2 to 50 g/L were used to establish the calibration curve. The calibration equation was y = 1.4088x and the coefficient of determination was 0.993.

Total Carbohydrate Content
Total carbohydrate content in UF permeate and hydrolyzed permeate was determined according to the anthrone-sulfuric acid method of E. W. Yemm and A. J. Wilis [48]. Glucose was used as standard and glucose solutions with concentrations ranging from 0.1 to 1 mg/mL were used to construct the calibration curve (y = 1.1845x, R 2 = 0.9981).

Protein Content
Protein content in UF permeate was assessed using the Kjeldahl method (AOAC, 1995) [49]. These compounds were low molar weight peptides or free amino acids since proteins were fully retained by the membrane. A nitrogen-to-protein factor of 6.25 was used as frequently used for rapeseed meal.

Resin Screenings
The resins were screened on the basis of massic adsorption capacity (q e , amount of phenolic compounds adsorbed per g of resin) calculated as: where C 0 and C e are the initial and equilibrium concentrations of phenolic compounds in non-hydrolyzed and hydrolyzed permeate solution, respectively (mg/mL); V i is the initial volume of permeate added onto the resins (mL); W is the weight of the dried resin (g).

Adsorption Kinetics
The adsorption capacity was monitored after 5, 10, 15, 30, 60, 90, and 120 min. To do so, phenolic compounds' concentration was measured in the liquid phase by HPLC. q e was deduced from the concentration. Results were plotted under linearized models (pseudo-first-order, pseudo-second-order, and intra-particle diffusion).

Adsorption Isotherms
Adsorption isotherms expressed the relationship between phenolic compounds' adsorption capacity (q e , mg/g) and the concentration of sample solution in the liquid phase at the equilibrium (C e , mg/L). For the adsorption study, a duration of 120 min was chosen. Experiments were carried out at 25 • C. Langmuir and Freundlich models were used to regress experimental data.

Adsorption Thermodynamic Parameters
The effect of the temperature was investigated by determining the adsorption isotherms at 298.15, 308.15, and 318.15 K. Enthalpy and entropy variations were obtained from the slope and intercept of the linear plot lnK eq vs. 1/T according to the linear form of Clausius-Clapeyron Equation (2): where lnK eq is the natural logarithm of the constant of adsorption equilibrium (K eq ), ∆H is the enthalpy change (J/ mol), ∆S is entropy change (J/mol), R is the universal gas constant (8.3144 J/molK), and T is the absolute temperature in Kelvin (K). ∆G was determined using Equation (3): where ∆G (J/mol) is the Gibbs energy change.

Desorption
Desorption ratios were determined using different water/ethanol solutions after the adsorption step on XAD16 up to equilibrium. Forty milliliters of 30, 50, 70, and 90% water/ethanol (v/v) was added to the resin and shaken at 150 rpm and 25 • C for 2 hours to reach desorption equilibrium. Resins were washed twice with deionized water prior to solvent addition. Resins were separated from the liquid filtration using filter paper. Phenolic compounds concentration in the liquid was quantified by HPLC. The desorption ratio was calculated using Equation (4): where C d is the concentration of CGA in desorption solution (mg/mL) and V d is the volume of the desorption solution (mL).
3.6. Antioxidant Activity 3.6.1. DPPH Radical Scavenging Assay DPPH free radical scavenging activity was carried on according to C. Wu et al. [50] with some modifications. Briefly, DPPH 0.2 mM in MeOH was prepared. one hundred microliters of samples at 1.25, 2.5, 5, 10, and 20 µg/mL was mixed with 100 µL of DPPH 0.2 mM and added to the wells of a microplate. The mixtures were kept at 25 • C for 30 min in the dark and then shaken for 30 s. The absorbance was measured at 517 nm. The inhibition percentage (%) of radical scavenging capacity was expressed as follows: where A DPPH is the absorbance of the DPPH solution, A blank is the absorbance of pure methanol, A sample+DPPH is the absorbance of DPPH with the sample, and A sample+blank is the absorbance of pure methanol with the sample.

ABTS Radical Scavenging Assay
The ABTS assay was used according to the procedure described by R. Re et al. [51] with some modifications. ABTS solution (3.5 mM) and potassium persulfate (1.225 mM) were mixed to produce the ABTS + solution. This solution was kept at room temperature in the dark for 16 h before use. When the radical had formed, the absorbance of ABTS + solution at 734 nm was adjusted to 0.7 ± 0.02 by dilution with 95% (v/v) methanol solution with a 1:32 ratio (v/v). Twenty microliters of samples at different concentrations (20, 10, 5, 2.5, and 1.25 µg/mL) was mixed with 180 µL ABTS + solution and added into each well of a 96-well plate. After 5 min of incubation in the dark at 25 • C, the plate was shaken for 20 seconds and the absorbance was measured at 734 nm. Ascorbic acid (vitamin C), SP, and SA were used as references. The following equation was used to calculate ABTS + inhibition rate (%): where A ABTS is the absorbance of ABTS + and A sample+ABTS is the absorbance of ABTS + with the sample.
All measurements for all sample solutions at each concentration were performed in triplicate.

Calculation of IC50
The IC50 (inhibitory concentration) value is the antioxidant concentration required to scavenge 50% of DPPH and ABTS free radicals. These values were calculated from the graph of radical scavenging activity against the different concentrations of tested samples. The concentration of the sample solution at IC50 (µg/mL) was determined by regression.

Data Analysis
Results are presented as the means ± S.D. (standard deviation) from three replicates of each experiment. All tests were considered significant at p < 0.05. Statistical analyses were performed using Rstudio 3.6.1 (Boston, MA, USA) open-source code. All figures were implemented by using the OriginPro 8.5 software (Northampton, MA, USA). The chemical structures of phenolic compounds were represented using the ChemDrawUltra 8.0 package (Cambridge Soft, MA, USA).

Conclusions
In conclusion, this study provides insights into the determination of phenolic compounds in an aqueous protein extraction/purification by-product (permeate) from rapeseed meal (RSM). We found that sinapine was the main phenolic compound in this permeate. The other compounds were sinapoyl glucose, 1,2-di-sinapoyl gentiobise, and sinapic acid. The acidic hydrolysis of this permeate allowed the efficient conversion of sinapine into sinapic acid. The XAD16 resin showed the highest adsorption capacity. Adsorption behaviors of total phenolic compounds, sinapine, and sinapic acid were also investigated. Our results revealed that the adsorption of phenolic compounds in both the hydrolyzed and non-hydrolyzed fractions followed the pseudo-second-order model and presented a very similar pattern of intra-particle diffusion. The Langmuir model was suitable to describe the adsorption process of phenolic compounds. Furthermore, the adsorption process was an exothermic, randomness, and physical adsorption process. Finally, we showed that the hydrolyzed fraction has the potential to become an interesting natural antioxidant, by comparison to vitamin C, that could find applications in food preservation or other domains.  Acknowledgments: This work was supported by "Impact Biomolecules" project of the "Lorraine Université d'Excellence", funded by the ANR "Investissements d'avenir" [grant number .

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
Sample Availability: Samples of the compounds are not available from the authors. Rate constant of the intra-particle diffusion model (mg/g×min0.5) C

Abbreviations
Constant associated with the thickness of the boundary layer (mg/g) Q max Maximum saturated monolayer adsorption capacity of polyphenol (mg/g) K L Langmuir constant (L/mg) K F Freundlich constant (mg/g)/(mg/L)n n Freundlich intensity parameter (0 < n ≤1) ∆H Enthalpy change (kJ/mol) ∆S Entropy change (kJ/mol K) ∆G Gibbs free energy change (kJ/mol K) R Ideal gas constant (8.314 J/mol K) T Temperature (K) K eq Equilibrium distribution coefficient of adsorption isotherm S.D.
Standard deviation