Hydroxypropyl- β -Cyclodextrin as a Green Co-Solvent in the Aqueous Extraction of Polyphenols from Waste Orange Peels

: There is, to-date, an expanding interest concerning the use of cyclodextrins as green food-grade co-solvents in the aqueous extraction of polyphenols, however, data regarding polyphenol extraction from waste orange peels (WOP) are lacking. On this ground, hydroxypropyl β -cyclodextrin (HP- β -CD), a highly water-soluble cyclodextrin, was used to develop a simple and straightforward methodology for the e ﬀ ective recovery of WOP polyphenols. Process optimization by response surface showed that maximum total polyphenol recovery (26.30 ± 1.49 mg gallic acid equivalents g − 1 dry mass) could be accomplished with 15 mM HP- β -CD at 40 ◦ C. On the other hand, integration of ultrasonication pretreatment was found unsuitable, as it resulted in reduced polyphenol yield. Examination of solvent acidity indicated that polyphenol extraction may be enhanced at pH 4, but the di ﬀ erence was non-signiﬁcant ( p > 0.05) compared to yields attained at pH 2, 3, and 5. Extraction of WOP polyphenols with HP- β -CD was shown to provide signiﬁcantly higher hesperidin yield compared to 60% ( v / v ) aqueous ethanol, which suggested selectivity of HP- β -CD toward this polyphenolic metabolite.


Introduction
Currently, agri-food industry waste represents a highly significant environmental issue and valorization of fruit and vegetable side streams is paramount to establishing sustainable routes of waste management. To-date, waste biomass originating from food industry processing residues is rather undervalorized, being used as a source of fertilizers, fuel, and livestock feeds. However, plant-food processing by-products and wastes may now be regarded as a prominent potential source for value-added natural products and chemicals [1,2]. At present, both emerging and conventional technologies must be combined in a coordinated manner to allow for an integral management of food processing wastes, thereby minimizing impacts on the environment, and reducing wastage of natural resources.
Globally, the production of citrus fruits for the year 2016/17 amounted to more than 8200 million metric tons, approximately 7209 of which was represented by oranges [3]. Citrus peel and pulp are the major waste streams of citrus juice processing, accounting for nearly 55-60% of the fresh fruit weight, and it has been estimated that the volume of such wastes may be more than 15 × 10 6 tons worldwide [4]. There is currently an expanding interest concerning the green valorization of citrus processing wastes for the recovery of a variety of value-added compounds such as flavonoids, carotenoids, pectins, and essential oil, which could be used in the pharmaceutical, food, and cosmetics industry as functional ingredients, additives, colorants, flavorings, etc. [5]. Flavonoids are a class of biopotent constituents

Waste Orange Peels
Orange (Citrus sinensis cv Valencia) peels were collected from an orange-processing plant (Prefecture of Thessaly, central Greece) and transported within 30 min to the laboratory, where they were processed with a sharp cutter to remove remaining flesh from peels (flavedo and albedo). This material, assigned as waste orange peels (WOP), was spread over stainless steel trays to layers of 0.5-cm thickness, and then dried in a laboratory oven (Binder BD56, Bohemia, NY, USA) for 480 min at 70 • C. After pulverization in a ball mill, the powder was sieved to obtain a uniform feed with an average particle diameter of 0.379 mm and stored at −17 • C.

Extraction Procedure
A precise amount of 0.667 g of WOP was transferred into a glass flask with 20 mL of deionized water (solvent-to-solid ratio 30 mL g −1 ) containing HP-β-CD. The mixture was subjected to ultrasonication and then extracted for 150 min at 500 rpm on a magnetic stirrer (Witeg, Wertheim, Germany). The concentration of HP-β-CD (C CD ), the extraction temperature (T), and the time of ultrasonication (t US ) were dictated by the experimental design. Ultrasonication was carried out in an Elma D-78224 Singen HTW heated ultrasonic bath (Elma Schmidbauer GmbH, Singen, Germany) and set at a power of 550 W, a frequency of 50 Hz, and acoustic energy density of 78.6 W L −1 . Ultrasound-assisted pretreatment of all samples was accomplished at ambient temperature (26 ± 2 • C). In the experiments carried out to assess the effect of pH, citric acid was added to the extraction medium at a standard concentration of 1 g L −1 . Then, the pH of the medium was adjusted by using 5 M NaOH. The extracts generated were centrifuged for 10 min at 10,000× g.

Experimental Design and Response Surface Optimization
The three process (independent) variables selected to set up the design of experiment were the concentration of HP-β-CD (C CD ), the time of ultrasonication pretreatment (t US ), and the temperature (T). Construction of the predictive model was performed by deploying a response surface methodology via a Box-Behnken design with three central points. The basis for choosing the process variables was literature data [15,16] illustrating their profound influence on the extraction yield in total polyphenols (Y TP ), which was the response. The actual and codified levels of the variables are analytically presented in Table 1. Codification of variable levels was accomplished as described elsewhere in detail [17]. Assessment of the model derived was based on the lack-of-fit test and analysis of variance (ANOVA). These tests enabled the determination of the significance of the model as well as of each polynomial coefficient and the model coefficient R 2 . The model (mathematical equations) was composed of only significant terms and insignificant ones (p > 0.05) were not considered.

Total Polyphenol Determination and Antioxidant Activity
The concentration of total polyphenols in the extracts produced was determined with a Folin-Ciocalteu methodology reported elsewhere [17] using gallic acid as calibration standard. Yield in total polyphenols (Y TP ), expressed as mg gallic acid equivalents g −1 dry mass (dm), was computed as previously described [18]. The antioxidant activity was estimated by performing antiradical activity (A AR ) and a ferric-reducing power (P R ) assays, as described earlier in detail [17]. Expression of the results was in µmol DPPH g −1 dm and µmol ascorbic acid equivalents (AAE) g −1 dm for A AR and P R , respectively.

Statistical Analysis
At least two extractions were performed for each experiment and analyses were carried out in triplicate. All values reported are means ± standard deviation (sd). SigmaPlot™ 12.5 (Systat Software Inc., San Jose, CA, USA) was employed to perform linear correlations (95% significance level). JMP™ Pro 13 (SAS, Cary, NC, USA) was employed to do distribution analysis, set up the experimental design, and run response surface statistics (ANOVA, lack-of-fit).

Design of Experiment and Extraction Process Optimization
The evidence emerged by recent studies on HP-β-CD-aided polyphenol extraction, revealed the importance of its concentration on the extraction performance [19][20][21]. Likewise, the time of ultrasonication pretreatment was also shown to be crucial in this regard [16,17]. Thus, these two parameters, along with temperature, whose effect on the extraction performance is undisputed, were selected to be the independent variables of the experimental design, and Y TP was chosen as the response. The influence exerted by each of the process (independent) variables was assessed by performing ANOVA, and the overall fitting of the predictive model to the experimental measurements was assessed by lack-of-fit and ANOVA ( Figure 1). design, and run response surface statistics (ANOVA, lack-of-fit).

Design of Experiment and Extraction Process Optimization
The evidence emerged by recent studies on HP-β-CD-aided polyphenol extraction, revealed the importance of its concentration on the extraction performance [19][20][21]. Likewise, the time of ultrasonication pretreatment was also shown to be crucial in this regard [16,17]. Thus, these two parameters, along with temperature, whose effect on the extraction performance is undisputed, were selected to be the independent variables of the experimental design, and YTP was chosen as the response. The influence exerted by each of the process (independent) variables was assessed by performing ANOVA, and the overall fitting of the predictive model to the experimental measurements was assessed by lack-of-fit and ANOVA ( Figure 1).
The non-significant terms ( Figure 1, "Parameter estimates") were excluded from the model, hence its final form was: The square correlation coefficient (R 2 ), which is an indicator of the total variability around the mean provided by the model [22], was 0.99 (p = 0.0001), and was determined by taking into consideration the proximity of the predicted and measured response values ( Figure 1B). These values are given in full detail in Table 2. This fact suggests that the model was very well adjusted to the experimental data. The visualization of the effect of independent variables on the response was based on the model derived, and it was presented in the form of 3D plots ( Figure 2). The desirability function ( Figure 1A) enabled the estimation of the maximum YTP, which was 26.30 ± 1.49 mg GAE g −1 dm. The optimal settings predicted by the model to achieve this value were CCD = 15 mM, tUS = 0 min, and T = 40 °C. To confirm the validity of this prediction, three individual extractions were carried out under these The non-significant terms ( Figure 1, "Parameter estimates") were excluded from the model, hence its final form was: The square correlation coefficient (R 2 ), which is an indicator of the total variability around the mean provided by the model [22], was 0.99 (p = 0.0001), and was determined by taking into consideration the proximity of the predicted and measured response values ( Figure 1B). These values are given in full detail in Table 2. This fact suggests that the model was very well adjusted to the experimental data.
The visualization of the effect of independent variables on the response was based on the model derived, and it was presented in the form of 3D plots ( Figure 2). The desirability function ( Figure 1A) enabled the estimation of the maximum Y TP , which was 26.30 ± 1.49 mg GAE g −1 dm. The optimal settings predicted by the model to achieve this value were C CD = 15 mM, t US = 0 min, and T = 40 • C. To confirm the validity of this prediction, three individual extractions were carried out under these

The Effect of pH Regulation
The influence of the pH of the extraction medium on YTP was assessed through regulation of the pH in deionized water and in 15 mM HP-β-CD solution by adding 1 g L −1 citric acid. Extraction was then carried out at 40 °C, and the results are depicted in Figure 3. Over the pH range of 2 to 5, the difference in YTP between deionized water and HP-β-CD was from 33.6% (pH 3) to 54% (pH 4), highlighting the effect of HP-β-CD on WOP polyphenol recovery. For the extraction with HP-β-CD, The maximum Y TP attained (26.30 ± 1.49 mg GAE g −1 dm) was comparable to 26.88 mg GAE g −1 [23] and 23.63 mg GAE g −1 dm [24], both achieved with microwave-assisted extraction, but far higher than the 7 mg GAE g −1 dm achieved with high voltage electrical discharges combined with enzymic hydrolysis [25], 7.29 mg GAE g −1 dm with aqueous extraction [26], 13.89 mg GAE g −1 dm with ultrasound-assisted extraction [27], and approximately 16 mg GAE g −1 dm achieved with infrared Beverages 2020, 6, 50 6 of 12 heating pretreatment and ultrasonication extraction [28]. Despite differences in the polyphenol peel content, which may arise from the influence of factors such as sun exposure, soil type, cultivation practices, maturation stage, etc. [24], critical comparison of the above data might indicate that the HP-β-CD-aided extraction was a high-performance process. The optimum C CD value of 15 mM was the highest level used in the experimental design, close to HP-β-CD solubility [29]. This finding suggests that increased Y TP may be achieved simply by raising C CD . Such a trend was also reported for the extraction of red grape polyphenols [30] and pomegranate fruit [19]. On the other hand, other investigations stressed that after a certain point, increases in C CD resulted in non-significant increases in Y TP [21]. This could enable the use of amounts lower of HP-β-CD, resulting in a lower-cost process. Therefore, a thorough examination of the effect of C CD on Y TP should be an indispensable element of an extraction process development.
The time of ultrasonication pretreatment (t US ) had a negative effect on WOP polyphenol extraction ( Figure 1, "Parameter estimates"), and according to the model (Equation (1)), maximization of Y TP required no ultrasonication pretreatment (t US = 0 min). This phenomenon might appear to be a paradox, considering the undisputed contribution of ultrasonication on the polyphenol extraction from peels of several citrus species [31][32][33][34][35]. Furthermore, the integration of ultrasonication as a pretreatment stage has been demonstrated to significantly enhance subsequent stirred-tank extraction [16]. However, data on the effect of ultrasonication on the cyclodextrin-aided polyphenol extraction are extremely limited, and therefore the actual effect is largely unknown. The combination of β-cyclodextrin with ultrasound-assisted extraction (UAE) for polyphenol recovery from some vinification wastes has been shown to provide exceptional boosting in extraction yield [36]. Contrary to that, extraction of polyphenols with various deep eutectic solvents (DES) in the presence of β-cyclodextrin was either enhanced or suppressed [37]. Although these discrepancies would not allow for argument on the effect of HP-β-CD/ultrasonication combination on WOP polyphenol extraction, it could be hypothesized that incorporation of HP-β-CD in the extraction medium during ultrasonication pretreatment might obstruct polyphenol extraction because of the increased viscosity of the liquid phase, which reduces solute (polyphenol) diffusivity [18].
Subsequent stirred-tank extraction is most probably favored by the presence of HP-β-CD, since stirring creates turbulence that fosters diffusivity [38] and enables the formation of polyphenol/HP-β-CD complexes, which enhance polyphenol solubility [29]. Based on the latter argument, it could also be supported that during ultrasonication pretreatment, irradiation with ultrasounds might destabilize polyphenol/HP-β-CD complexes and does not contribute to polyphenol solubilization in the liquid phase. This could be another reason for the negative effect of the ultrasonication pretreatment observed.
The effect of temperature (T) was also negative, and the optimum level determined was 40 • C, which represented the lowest limit tested (Table 1). This outcome was in line with a recent study on HP-β-CD-aided polyphenol extraction from potato peels [21], where it was found that T > 30 • C did not favor increased polyphenol yield. Increases in T have been explicitly demonstrated to increase the aqueous solubility of sparingly soluble polyphenols in the presence of various cyclodextrins [39]. On the other hand, it was found that the binding constants of certain polyphenol complexes with α-, β-, and γ-cyclodextrin decreased as T increased [40]. Results on the interactions of the stilbene (E)-piceatannol with HP-β-CD were in concurrence, illustrating a decrease in the complexation constant as a response to increasing T [41]. Such a phenomenon was attributed to weakening of hydrogen bonds between the guest and the host molecule due to heating. In addition, it was observed that at around 50-60 • C, there may be decomposition of polyphenol-cyclodextrin inclusion complexes, and therefore the reduced Y TP found for T > 40 • C can be justified.

The Effect of pH Regulation
The influence of the pH of the extraction medium on Y TP was assessed through regulation of the pH in deionized water and in 15 mM HP-β-CD solution by adding 1 g L −1 citric acid. Extraction was then carried out at 40 • C, and the results are depicted in Figure 3. Over the pH range of 2 to 5, the difference in Y TP between deionized water and HP-β-CD was from 33.6% (pH 3) to 54% (pH 4), highlighting the effect of HP-β-CD on WOP polyphenol recovery. For the extraction with HP-β-CD, the highest Y TP was found at pH 4 (29.13 ± 1.10 mg GAE g −1 dm). However, no statistical difference was seen compared to the Y TP achieved at the other pH tested. This is in contrast to recent studies on the importance of the pH in the HP-β-CD-aided extraction of polyphenols [15,16], where pH effects on polyphenol extraction performance was found to be significant. Optimum pH value for HP-β-CD-aided extraction of S. fruticosa polyphenols was shown to be 3.90 [16], very close to pH 4 determined in this study.
Beverages 2020, 6, x FOR PEER REVIEW 9 of 13 the highest YTP was found at pH 4 (29.13 ± 1.10 mg GAE g −1 dm). However, no statistical difference was seen compared to the YTP achieved at the other pH tested. This is in contrast to recent studies on the importance of the pH in the HP-β-CD-aided extraction of polyphenols [15,16], where pH effects on polyphenol extraction performance was found to be significant. Optimum pH value for HP-β-CDaided extraction of S. fruticosa polyphenols was shown to be 3.90 [16], very close to pH 4 determined in this study. Since pH affects the dissociation of phenolic hydroxyls and carboxyl functions, its role in polyphenol-cyclodextrin interactions has been implicated. Investigations on caffeic acid (CA)/HP-β-CD interactions illustrated that complex formation constants were sensitive to pH, manifesting a decline with pH increases [42]. Such a phenomenon was attributed to CA hydrophobicity, which is Since pH affects the dissociation of phenolic hydroxyls and carboxyl functions, its role in polyphenol-cyclodextrin interactions has been implicated. Investigations on caffeic acid (CA)/HP-β-CD interactions illustrated that complex formation constants were sensitive to pH, manifesting a decline with pH increases [42]. Such a phenomenon was attributed to CA hydrophobicity, which is related to its ionization state. Considering that hydrophobic interactions between the guest molecule and the HP-β-CD cavity contribute to the formation of inclusion complexes, acidic pH would maintain CA in a non-ionized and therefore more hydrophobic form. This would entail greater stability constant. Data on rosmarinic acid/β-CD interactions were along the same line [43]. A similar conclusion was reached for catechin and epicatechin interactions with HP-β-CD [44].

Comparative Assessment and Polyphenolic Composition
To further illuminate the potential of HP-β-CD as a co-solvent in the aqueous extraction of WOP polyphenols, a comparative appraisal was performed, considering not only Y TP , but also the yield in total flavonoids (Y TFn ), the antiradical activity (A AR ), and the ferric-reducing power (P R ) ( Table 3). Table 3. Appraisal of hydroxypropyl β-cyclodextrin (HP-β-CD) as a co-solvent in the production of polyphenol-containing aqueous extracts from waste orange peels (WOP). Extractions were performed at 40 • C, and values given represent average ± standard deviation. Compared to 60% ethanol, which was the most effective conventional solvent, extraction with HP-β-CD gave by 30% higher Y TP , virtually equal Y TFn , 7.5% lower A AR (p > 0.05), and 10.9% lower P R (p > 0.05). This outcome demonstrated that HP-β-CD was highly effective in assisting aqueous extraction of WOP polyphenols at 40 • C.

Solvent
The extracts obtained with HP-β-CD and 60% EtOH were analyzed by HPLC to identify the principal polyphenolic constituents and determine quantitative differences. The chromatogram recorded at 280 nm revealed that both extracts were dominated by two polyphenols, narirutin and hesperetin ( Figure 4). This was in accordance with earlier examinations, which found these two phytochemicals to be the major WOP metabolites [45][46][47]. The difference in narirutin recovery between HP-β-CD and 60% EtOH extraction was 10.4%, but the difference for hesperidin was 62% (Table 4). This fact indicates that there might be selectivity displayed by HP-β-CD toward hesperidin. Such a phenomenon would appear reasonable, since as pointed out above, less polar compounds interact more strongly with the HP-β-CD cavity. Thus, hesperidin, which is less polar that narirutin, may form more stable complexes with HP-β-CD that facilitate its solubilization and recovery.  16.20 phytochemicals to be the major WOP metabolites [45][46][47]. The difference in narirutin recovery between HP-β-CD and 60% EtOH extraction was 10.4%, but the difference for hesperidin was 62% (Table 4). This fact indicates that there might be selectivity displayed by HP-β-CD toward hesperidin. Such a phenomenon would appear reasonable, since as pointed out above, less polar compounds interact more strongly with the HP-β-CD cavity. Thus, hesperidin, which is less polar that narirutin, may form more stable complexes with HP-β-CD that facilitate its solubilization and recovery. Overall, HP-β-CD-aided extraction afforded a 54.6% higher yield than 60% aqueous EtOH. This finding demonstrates the high extraction efficiency of HP-β-CD. The level of narirutin determined (5.11 ± 0.20 mg g −1 dm) was in accordance with previous studies that reported values of 0.26-4.52 mg g −1 dm [46]. However, other authors reported narirutin Overall, HP-β-CD-aided extraction afforded a 54.6% higher yield than 60% aqueous EtOH. This finding demonstrates the high extraction efficiency of HP-β-CD.

Conclusions
This study illustrated the efficiency of WOP polyphenol extraction at low temperature by employing HP-β-CD. Such an extraction technique for WOP polyphenol recovery, to the best of the authors' knowledge, has not been previously reported. The results indicated that ultrasonication pretreatment provided no benefit to increasing extraction yield, and that extraction was favored at 40 • C. Under optimal settings, HP-β-CD-aided extraction produced extracts with total polyphenol yield and antioxidant characteristics comparable to those obtained with common conventional solvents such as aqueous methanol and ethanol. The principal polyphenolic phytochemicals identified in the extracts were narirutin and hesperidin. Future examinations should demonstrate the applicability of such extracts in food and cosmetics to pave the way for large-scale valorization of WOP as cost-effective, abundant, and bio-potent natural constituents.