and acidified milk gels as a strategy for improving stability and antioxidant activity of yarrow phenolic compounds after gastrointestinal digestion

The aim of this study was to improve the stability and antioxidant activity of yarrow phenolic compounds upon an in vitro simulated gastrointestinal digestion. Therefore, two types of caseins-based delivery systems, sodium caseinate stabilized nanoemulsions (NEs) and glucono delta-lactone acidified milk gels (MGs), were formulated containing an ultrasound-assisted yarrow extract (YE) at two concentrations (1 and 2.5 mg/mL). Formulations with 1 mg/mL of YE were chosen based on their higher encapsulation efficiency to perform the in vitro digestion experiments. After digestion, YE-loaded NEs only partially protected phenolic compounds from degradation; meanwhile the phenolic composition of YE including in MGs after digestion was quite similar to undigested YE. Moreover, the antioxidant activity of MGs after digestion was higher than NEs digested samples, which confirms the higher protection of YE phenolic compound by the milk gels systems. This research demonstrated the potential use of acidified MGs as carriers to improve the stability and antioxidant activity of yarrow phenolic compounds. Therefore, these matrices could be employed to develop new dairy products enriched with phenolic compounds.


Introduction
Achillea millefolium L., commonly known as yarrow, has been traditionally used to treat gastrointestinal and hepatobiliary disorders, inflammation or diabetes (Akram, 2013).The bioactivity of aqueous and ethanolic yarrow extracts has been associated with the presence of phenolic compounds, mainly phenolic acids and flavones (Villalva et al., 2019 ).Nevertheless, after oral consumption, the potential health benefits of phenolics compounds could be limited due to instability under certain pH conditions, poor water solubility, gastrointestinal enzyme degradation, low intestinal permeability or metabolism during digestion (Karakaya, 2004 ).
Thus, phenolic stability during gastrointestinal digestion has been related to their chemical structure; for instance, the presence or absence of glycosylation, the glycosylation patterns, etc. (Czubinski et al., 2019).Besides, alkaline medium of small intestine has been also described as an important parameter in phenolic compounds stability reduction (Chen et al., 2014).
Moreover, digestive enzymes were reported to accelerate the hydroxylation of phenolics after gastric digestion phase (Barak, Celep, Inan & Yesilada, 2019).Therefore, oral administration of these compounds requires strategies to preserve their chemical integrity and successful delivery in physiological targets.
Ideally, food-grade encapsulation materials, as proteins, are selected to formulate food products.Proteins have multiple applications in encapsulation matrices of phenolic compounds, as emulsifiers, gelling agents and also as entities to formulate micro and nano-particles (Jia, Dumont & Orsat, 2016;Foegeding, Plundrich, Schneider, Campell & Lila, 2017).The ability of phenolic compounds to associate with milk proteins is well documented, particularly to proline-rich proteins such as αand β-caseins through hydrophobic interactions and hydrogen bonds (Chanphai, Bourassa, Kanakis, Tarantilis & Polissiou, 2018).Thus, this ability has been used to develop delivery carriers.Ghayour at al. (2019) proposed the nanoencapsulation of quercetin and curcumin by using caseins in order to design protein-based delivery systems to incorporate these compounds to beverages.Arranz et al. (2019) also employed casein matrices as carriers for rosmarinic acid delivery in food product development.. Nanoemulsion-based encapsulation systems for phenolic compounds have been well established, due to its high efficacy encapsulation, maintenance of chemical stability and controlled release (Artiga-Artigas, Lanjari-Pérez & Martín-Belloso, 2018;Lu, Kelly & Miao, 2016).In order to develop these systems, sodium caseinate has been described as a food grade emulsifier that improves delivery of phenolic compounds.Sabouri, Arranz, Guri & Corredig (2018) indicated that the incorporation of epigallocatechin-gallate (EGCG) in sodium caseinate emulsions improved its bioaccessibility and preserved its functional properties during gastrointestinal digestion.Besides, Casanova et al. (2018) also reported a complexation of cyaniding-3-O-glucoside with sodium caseinate, suggesting that sodium caseinate is a potential carrier for anthocyanins.
The gelation of protein matrices has also been indicated as an alternative method to encapsulate phenolic compounds (Jia et al., 2016).Caseins have excellent gelation properties and have been proposed as carriers for phenolic compounds since they are able to interact with phenolics, form complexes and entrap them through their gelation process (Ozdal, Capanoglu & Alta, 2013).Thus, Bayraktar, Harbourne & Fagan (2019) showed that polyphenols extracts (green tea and white grape) and simple phenolic compounds (tannic and gallic acids) could be successfully used to develop acidified dairy products with antioxidant properties.Moreover, Trigueros, Wojdylo & Sendra (2014) also reported that in yogurts containing pomegranate juice, most of anthocyanins were bound to casein fraction.
The aim of this study was first to evaluate the stability of yarrow phenolic compounds after in vitro simulated gastrointestinal digestion.Then, two different casein-based delivery systems were developed and characterized; as a strategy to improve yarrow phenolic compounds stability and antioxidant activity after simulated digestion.
The sample included inflorescences and upper dried leaves of the plant (water content less than 5%).The plant was ground using a Premill 250 hammer mill (Leal S.A., Granollers, Spain) and sieved to the appropriate size (<500 µm).The extraction process was performed by UAE in an ultrasonic device (Branson digital sonifier 450, Danbury, USA) at 400 W and 60 kHz.Ethanol was used as extraction solvent (1:10 plant/solvent ratio) during 30 min at 40 ºC and an output of 50% nominal amplitude.Ethanol was evaporated in a rotary evaporator (IKA ® , Werke GmbH and Co., KG, Germany) and samples were stored at −20ºC until use.

Chemical characterization of phenolic compounds
Phenolic chromatographic analyses were conducted in an Agilent 1260 Infinity series HPLC-PAD (Santa Clara, CA, USA).A reverse phase ACE Excel 3 Super C18 column (150 mm x 4.6 mm, 3 m particle size) from ATC (Aberdeen, Scotland, UK) equipped with a pre-column from the same material was employed according to the chromatographic method reported by Villalva et al. (2018).Besides, accurate mass from HPLC-PAD-QTOF-MS/MS in a negative mode analysis was used for compounds identification (Villalva et al., 2018).Quantification of identified compounds was performed using calibration curves of pure analytical standards.In addition, luteolin-di-hexoside and 6-hydroxyluteolin-7-O-glucoside were quantified by the calibration curves of orientin and luteolin-7-O-glucoside respectively.Correspondingly, apigenin glycosylated derivative were quantified using apigenin-7-O-glucoside calibration curve and schaftoside isomer and apigenin-hexoside-pentoside respect to schaftoside calibration curve.
NEs were prepared using high pressure homogenization (HPH).Oil in water emulsions, containing 7% of soybean oil (w/w) (Sigma Aldrich, Oakville, ON, Canada) and a YE stock solution (dissolved in less than 1% of EtOH) to achieve 0, 1 or 2.5 mg/mL in final emulsion, were formulated with 2% sodium caseinate (NaCas) (New Zealand Milk Proteins, Mississauga, ON, Canada) as emulsifier.Initially, NaCas was dispersed in ultrapure water with magnetic stirrer and stored overnight at 4ºC to allow complete hydration.Emulsions were prepared by addition of YE to soybean oil and subsequent incorporation of NaCas solution.Both phases were pre-homogenized using a Polytron mixer (Brinkmann Inst.Corp., Mississauga, ON, Canada) at 30,000 RPM for 2 min, and then HPH was performed at 450 kPa for five passes using a microfluidizer (model M-110Y, Microfluidics Corporation, Newton, MA, USA).Then NEs were characterized (particle size and -potential) and stored under refrigeration at 4ºC for two weeks in order to evaluate its stability.
The particle size distribution was measured using static light scattering (Mastersizer 2000, Malvern Instruments, Worcestershire, UK), with water as dispersant.The refractive indices were 1.33 and 1.47, for water and soybean oil, respectively.Mean particle diameter was reported as the volume-weighted mean diameter (d4,3) calculated as the average of triplicate measurements.Zeta ()-potential was determined by dynamic light scattering (DLS) (Zetasizer Nano, Malvern Instruments, Malvern, UK) in samples diluted in ultrapure grade water at 1:500 ratio.All measures were done by triplicate.

Preparation and characterization of milk gels (MGs) of yarrow extract
Pasteurised skimmed milk, acidified with 2% (w/w) of glucono delta-lactone (GDL) (Sigma Aldrich) and containing a YE stock solution to obtain 0, 1 or 2.5 mg/mL in final gels, was used to prepare MGs.Initially, a suspension of heated (85ºC for 30 min) or non-heated milk with YE was stirred for 1 hour at room temperature and protected from direct light.Then, this suspension was warmed up to 30ºC, the GDL was added, and the mixture was kept at 30ºC for 180 min.
After milk gelation, the samples were stored at 4ºC until analyses.
The dynamic rheological measurements were performed by applying a constant strain of 0.05% at a frequency of 1 Hz by means of a Physica MCR301 stress-controlled rheometer (Anton-Paar, Graz, Austria) using a concentric cylinder with solvent trap cover, at 30°C.The gelation point was taken at the time of cross-over between Gʹ (storage modulus) and Gʺ (loss modulus) (Haratifar & Corredig, 2014).In parallel, the pH was monitored during milk gel formation using an Accumet pH meter (Fisher Scientific, Edmonton, Canada).All experiments were done by triplicate.

Encapsulation efficiency of phenolic compounds
The encapsulation efficiency (EE) of yarrow phenolic compounds in NEs and MGs was measured after centrifugation of fresh samples for 1 h (60.000 g (NEs) and 20.500 g (MGs), respectively).Aqueous phase were recovered and filtered using 0.22 m polyvinylidene fluoride (PVDF) filters prior injection in a HPLC-PAD.Analysis was performed as previously described in section 2.3.
Encapsulation efficiency of individual phenolic compounds (individual EE%) was determined by Eq. 1 and the total encapsulation efficiency (total EE%) of the formulation was calculated by Eq. 2, as followed: Eq. ( 1) Eq. ( 2) 2.7.In vitro simulated gastrointestinal digestion A two steps in vitro simulated digestion, gastric and intestinal, was performed according to a standardised protocol (Minekus et al., 2014), with slight modifications, suited to the food matrix being analysed (Arranz et al., 2017).Briefly, 10 mL of each sample were mixed with 10 mL of simulated gastric fluids containing 25,000 U/mL of pepsin, 0.2 mL of HCL (1 M) and 5 µL of CaCl2 (0.3 M) at a final pH 3 in amber jars.Samples were incubated in a shaking water bath (220 strokes/min) (New Brunswick Scientific Co., Inc., NJ, USA) for 30 min at 37°C.Then, 20 mL of simulated intestinal fluids were incorporated containing 5.0 mg/mL of pancreatin (800 U/mL), 160 mM bile salts, 40 µL of CaCl2, (0.3 M), 0.15 ml of NaOH (1 M), 1 mM phospholipids and 5 µL phospholipase A2 (stock solution 6.7 mg/mL), and the mixture was incubated for 2 h at 37 °C and pH 7. At the end of the digestion, the enzyme reaction was immediately stopped by keeping the samples on ice and stored at −20 °C.Samples (4 mL) were then freeze-dried for further experiments.Digestion process was performed by triplicate for every formulation (free YE, YE-NEs and YE-MGs).In addition, control digestions without YE (water, NEs and MGs) were performed.
For quantification of phenolic compounds after digestion, digested freeze-dried samples were re-suspended in 1 mL of EtOH and acidified water with 0.1% formic acid (60:40, v/v).Later, the mixture was placed into an ultrasound water-bath (5 min) and centrifuged at room temperature (4,500 RPM, 5 min).The clear supernatant was filtered using 0.45 m PVDF filters prior analysis by HPLC-PAD as described in 2.3 section.

Effect of in vitro simulated gastrointestinal digestion on YE phenolic compounds
Phenolic compounds characterization of YE, before and after in vitro simulated gastrointestinal digestion, was performed by HPLC-PAD analysis (Table 1).Therefore, twenty-one phenolic compounds were identified in YE, mainly glycosylated flavonoids and phenolic acids.The main compounds were 3,5-dicaffeoylquinic acid (DCQA) and luteolin-7-O-glucoside, followed by vicenin 2 and an apigenin glycosylated derivate.These phenols represented 49% of the total phenolic compounds quantified in YE.Other caffeoylquinic acid derivatives like chlorogenic acid and 4,5-DCQA were found in representative amounts within the phenolic acids in YE.
The quantitative analysis of phenolic fraction of YE after digestion showed that the concentration of almost all detected compounds was modified.Several authors indicated that the stability of phenolic compounds during the gastrointestinal digestion process is strongly influenced by their chemical structure, since phenolics have a different sensitivity to pH variations and digestive enzymes activity (Goulas & Hadjisolomou, 2019;Lima et al., 2019).
These results suggest an isomerization of 3,5-DCQA towards 3,4-DCQA and 4,5-DCQA in pH conditions occurring during digestion.Similarly, D'Antuono, Garbetta, Linsalata, Minervini & Cardinali (2015) also detected an isomerization effect after gastrointestinal digestion of 3,5-DCQA (pure individual compound) with the presence of 4,5-and 3,4-DCQAs after intestinal step.Luteolin-7-O-glucoside, the second most abundant compound in YE, was also reduced in approx.25%, after digestion.This reduction has been generally attributed to the instability of flavonoid glycosides under pH values of small intestine, where several modifications such as oxidation and polymerization might occur (Celep, Akyüz, Inan & Yesilada, 2018).Therefore, the encapsulation of YE phenolic compounds in delivery systems could be an useful strategy to improve their stability during gastrointestinal digestion.

Formulation and characterization of YE-loaded nanoemulsions (NEs)
YE-loaded NEs with two different concentrations of YE, 1mg/mL (NE-1.0)and 2.5 mg/mL (NE-2.5),were formulated in soy oil in NaCas water emulsions.NEs control, without YE, was also formulated.
The particle size distribution (Fig. 1) indicated that the control emulsion had a monomodal distribution.This distribution did not change after YE incorporation, although NE-2.5 showed a wider size distribution than control and NE-1.0.Freshly prepared control emulsion (day 0) had a mean particle diameter (d4,3) of 248 ± 1.0 nm (Table 2).The addition of YE did not modify the emulsions particle diameter.All NEs had a negative net surface charge (-potential), with no statistical differences between the control and YE-loaded NEs.
The physical stability of YE-loaded NEs was monitored during two weeks.The values of mean particle diameter and -potential after 7 and 14 days of storage at 4ºC are shown in Table 2.
After 14 days, no statistical differences were observed neither in the mean particle diameter nor in the -potential of any formulation.These data denote the stability of YE-loaded NEs.Thus, caseinate stabilized nanoemulsions containing tea polyphenols (epigallocatechin-gallate) showed a mean particle diameter and -potential values quite similar to those found in this study (Sabouri et al., 2015).Besides, Sharma et al. (2017) also reported that a clove nanoemulsion stabilized with sodium caseinate presented a similar mean particle diameter and -potential.
This emulsion was found to be stable for 20 days.

Formulation and characterization of acidified milk gels (MGs) including YE
Milk gels including two different concentrations of YE, 1mg/mL (MG-1.0)and 2.5 mg/mL (MG-2.5),were prepared using heated or unheated milk, acidified with GDL.Control MGs (without YE) were also formulated.
MGs formulated with heated milk showed a shorter gelation time and higher pH values at gelation time than those obtained with unheated milk (Table 3).Heated MGs also showed higher final storage modulus (Gʹ) values than those formulated with unheated milk (Table 3), indicating that heated gels presented a higher stiffness at the end of the acidification process.
Besides, the final loss tangent values (tan δ were significantly lower for heated gels.At moderate heating temperatures (70-90ºC), whey proteins unfold and associate with casein micelles.During acidification, these unfolded-associated proteins act as a cross-link between protein particles and increase the number (and strength) of bonds in the gel network, consequently higher Gʹ values are obtained, pH increases at gelation and shorter gelation times are required (Donato, Alexander & Dalgleish, 2007).
Despite the differences found in the rheological properties between heated and unheated formulations, the addition of different concentration of YE did not significantly modify the gelation time, the pH at gelation time, the final G´ value or the final tan δ value (Table 3).
Similarly, Vega & Grover (2011) formulated acidified MGs including a commercial cocoa extract, rich in flavanols, showing that final Gʹ values and pH values at gelation time were not modified with extract addition, compared to the control gel.However, the addition of tannic and gallic acids to milk have resulted in faster gelation times and gels with Gʹ values higher than control samples (Harbourne, Jacquier & O'Riordan, 2011).These different results could be explained by interactions between milk proteins and phenolic compounds and it was described that these interactions are highly affected by the concentration, type and structure of phenolic compounds employed (Ozdal et al., 2013).

Encapsulation efficiency (EE) of YE phenolic compounds in NEs and MGs.
The encapsulation efficiency of YE individual phenolic compounds in NEs and MGs is shown in Table 4. Regarding NEs, the encapsulation efficiency was related to the YE concentration, NE-1.0 showed higher EE for all the compounds than NE-2.5.Encapsulation efficiency of all phenolic compounds in NE-1.0 ranged from 88.1 to 100%.However, in NE-2.5 the values were under 55%, except for luteolin, apigenin and diosmetin.Hence, the total encapsulation efficiency calculated in NE-1.0 was significantly higher than in NE-2.5 (92.3 ± 4.6% vs. 33.7 ± 0.9% respectively).These results could be explained since at a concentration of 1 mg/mL, most YE phenolic compounds were binding sodium caseinate at the oil-water interface.
However, the concentration of 2.5 mg/mL could saturated interaction sites in protein chain, due to the polyphenol/protein ratio was too high.Consequently, a high quantity of phenolic compounds was non-encapsulated and presented in the aqueous phase.
Regarding MGs, the encapsulation efficiency was also affected by the concentration of YE, but in a lesser extent than in NEs.Thus, MG-1.0 had a total encapsulation efficiency of 72.0 ± 2.7%, meanwhile MG-2.5 showed a 63.1 ± 2.6%.In MGs formulations, apigenin and luteolin glycosylated forms presented a lower EE than their aglycone forms.Thus, Xiao et al. (2011) also indicated that the glycosylation of flavonoids lowered their affinity for milk proteins.These authors endorsed the decrease of affinity after glycosylation to the non-planar structure, since molecules with near-planar structure can easily enter the hydrophobic pockets in proteins.
However, this decrease in EE of glycosylated forms could be also related to their lower hydrophobicity, with respect to non-glycosylated forms.Among phenolic acids, the DCQAs showed higher encapsulation efficiency than chlorogenic and caffeic acids.DCQAs higher encapsulation could be related to the higher hydrophobicity of DCQAs (with respect to chlorogenic and caffeic acids), that induces a stronger association between proteins and polyphenols (Adrar, Madani & Adrar, 2019).Nevertheless, their different structures (size, length and flexibility) could also affect protein-phenolic compounds interactions.

Effect of in vitro digestion on YE phenolic compounds included NEs and MGs.
Formulations with 1 mg/mL of YE (NE-1.0 and MG-1.0) were chosen based on their higher EE to perform in vitro digestion experiments.The quantitative analysis of individual phenolic compounds was performed after gastrointestinal digestion (Table 1).The quantification of phenolic compounds after NE-1.0 digestion showed that the digestion process modifies their concentration, but in a lower extent than when free yarrow extract was digested.Moreover, 10 of the 21 phenolic compounds detected in digested NE-1.0 decreased their concentration (compared to undigested YE), meanwhile after free YE digestion the quantity of were reduced.Regarding 3,5-DCQA, the most abundant compound in YE, a 3.4 times decreased was observed when free YE was digested versus 1.6 times reduction with NE-1.0.
Luteolin-7-O-glucoside's concentration was only reduced from 7.57 ± 0.17 to 6.51 ± 0.10 mg/g of sample when NE-1.0 was employed.These results suggested that some encapsulated compounds in NEs were partially released and exposed to digestion conditions, indicating that the formulation of YE-loaded NEs only partially protected the YE phenolic compounds during gastrointestinal transit.The effect of gastric digestion on NaCas stabilized emulsions was previously reported.Li, Ye, Lee, & Singh (2012) described that caseinate-stabilized emulsions flocculate after digestion at acidic conditions (pH 1-3) due to pepsin hydrolysis over interfacial proteins in the droplets surface.Nevertheless, Sabouri et al. (2018) reported a higher recovery of EGCG (but not 100%) after digestion, when it was formulated with NaCas stabilized emulsions compared to water solutions.They concluded that the complexation of EGCG and NaCas at the interface of emulsions protected EGCG from digestive degradation.
These findings are in line with the protection effect of YE phenolic compounds against degradation during digestion by MGs.Indeed, Lamothe, Azimy, Bazinet, Couillard & Britten (2014) described that the addition of a green tea extract to a yogurt matrix significantly decreased the rate and extent of proteolysis in the gastric phase.They suggested that the formation of complexes between milk proteins and green tea polyphenols increase protein stability during pepsin digestion.However, they also reported that the presence of tea extract did not influence protein hydrolysis in the intestinal phase, with the release of 100% of the polyphenols initially added.They concluded that interactions between green tea polyphenols and milk proteins help to maintain the integrity of polyphenols during digestion.Recently, Adrar et al. (2019) reported that polyphenols-proteins interactions would play a certain role in the preservation of polyphenols structure and function through the gastrointestinal tract transit.
Thus, MGs offered an almost complete protection of phenolic compounds during gastrointestinal digestion; meanwhile this protection is only partial when employing NEs.

Effect of in vitro digestion on free and encapsulated YE antioxidant activity
The antioxidant activity of undigested YE and digested samples: free YE, NE-1.0 and MG-1.0 was evaluated (Table 5).Empty (without YE) NEs and MGs were also digested and its antioxidant activity was determined (0.050±0.002 and 0.078±0.003mmol Trolox/g sample respectively).The values obtained with these empty formulations were subtracted from the formulations with the extract.The digestion process reduced almost 50% of the antioxidant activity of free YE samples, which could be related with the reduction in phenolic compounds found (Table 1).The gastrointestinal digestion of NE-1.0 also decreased the antioxidant activity of the samples, however, in a lesser extent than free YE solutions.These results agree with the partial protection provided by the NE to YE phenolic compounds.Moreover, the antioxidant activity of MG-1.0 after digestion was the highest compared to free YE and NE-1.0, which confirms the higher protection of YE phenolic compounds by the milk gels systems.Thus, several authors previously reported that protein-polyphenol complexes formation preserved polyphenols bioactivities, including antioxidant activity (Foegeding, et al, 2017).
DPPH • methanol solution was used.The reaction was started by adding 25 µL of sample or blank.The bleaching of DPPH • was followed at 517 nm at room temperature until the reaction was completed (3 h).All the samples were prepared at 4 different concentrations, by triplicate, to obtain the inhibition curve of DPPH • .The results were calculated as mmol Trolox equivalent/g extract (TEAC value).Analysis was done by triplicate.
2.9.Statistical analysisFor each sample, experiments were performed at least by triplicate and expressed as mean values and standard deviation.A one-way analysis of variance (ANOVA) followed by LSD test (Fisher's least significant difference) post-test were used to discriminated significant differences between means at p <0.05.Analysis was performed using Statgraphic v. Centurion XVI package for Windows (Statpoint Inc., Warrenton, VA, USA).

Table 3 .
Properties of acidified skim milk gels formulated with heated and unheated milk.

Table 4 .
Encapsulation efficiency (EE) of individual yarrow phenolic compounds presented in yarrow extract (YE).
a,bDifferent lower case letters mean statistical difference within nanoemulsions.A,B Different super case letters mean statistical difference within milk gels.Significance level at p<0.05 with Fisher's Least Significant Difference (LSD) test.