Food Research International 115 (2019) 319–327

Contents lists available at ScienceDirect

Food Research International

journal homepage: www.elsevier.com/locate/foodres

Impact of conventional/non-conventional extraction methods on the T

untargeted phenolic profile of Moringa oleifera leaves
⁎
Gabriele Rocchettia,1, Francesca Blasib,1, Domenico Montesanob, , Silvia Ghisonia,
⁎
Maria Carla Marcotullioc, Stefano Sabatinic, Lina Cossignanib, , Luigi Lucinia
a
Department for Sustainable Food Process, Università Cattolica del Sacro Cuore, Via Emilia Parmense 84, 29122 Piacenza, Italy
b
Department of Pharmaceutical Sciences, Section of Food Science and Nutrition, University of Perugia, Via San Costanzo 1, 06126 Perugia, Italy
c
Department of Pharmaceutical Sciences, Section of Chemistry and Technology of the drug, University of Perugia, Via del Liceo 1, 06123 Perugia, Italy

A R T I C LE I N FO A B S T R A C T

Keywords: The impact of different extraction methods, namely maceration, homogenizer-assisted extraction, rapid solid-
UHPLC-ESI/QTOF liquid dynamic extraction, microwave-assisted extraction and ultrasound-assisted extraction, on polyphenols of
Extraction technologies Moringa oleifera leaves was studied. The phenolic composition of alcoholic (methanol 100%) and hydroalcoholic
Metabolomics (methanol/water 50:50, v/v) extracts was compared by using an untargeted metabolomics-based profiling ap-
Polyphenols
proach followed by multivariate statistics. With this aim, ultra-high-pressure liquid chromatography coupled to a
Antioxidants
quadrupole-time-of-flight mass spectrometry was used to profile phenolic compounds under the different ex-
traction conditions. Besides, the in vitro antioxidant activities of Moringa leaves were also investigated as ferric
reducing antioxidant power (FRAP) and oxygen radical absorbance capacity (ORAC). The metabolomic ap-
proach allowed to putatively annotate 262 phenolic compounds. In particular, glycosidic forms of quercetin (i.e.,
quercetin 3-O-galactoside, quercetin 3-O-glucoside, and quercetin 4’-O-glucoside) were the most represented
compounds among flavonoids. Furthermore, protocatechuic acid was found to be the most abundant hydro-
xybenzaldheyde derivative, while the isomeric forms of hydroxybenzoic acid characterized the phenolic acids
class. Overall, the extractions in methanol 100% were found to be the most effective for phenolic compounds
recovering, when compared with those in methanol/water (50:50, v/v). Homogenizer-assisted extraction of M.
oleifera leaves using 100% methanol allowed extracting the highest amounts of polyphenols (35.19 mg/g) and
produced the highest oxygen radical absorbance capacity (536.27 μmol Trolox Equivalents/g). The supervised
orthogonal projection to latent structures discriminant analysis identified phenolic acids as the phenolic class
mostly affected by the different extraction technologies. These findings demonstrate that each extraction method
promoted the recovery of specific phenolic subclasses with different efficiencies.

1. Introduction beneficial properties, being used as nutritional, medicinal and water

purification agent (Saucedo-Pompa et al., 2018).
Moringa oleifera Lam. belongs to the Moringaceae family and, to Several studies have demonstrated that the bioactive components
date, it represents one of the most important traditional multipurpose characterizing Moringa could be used for different industrial and food
food plants (Anwar, Ashraf, & Gilani, 2007). Moringa plants originate applications (Oyeyinka & Oyeyinka, 2018). In this regard, different
from India and Africa and are widely distributed in tropical and arid Moringa parts (i.e., seeds, stems, flowers and leaves) are characterized
countries (Saucedo-Pompa et al., 2018). Moringa plant is totally edible by a wide distribution of bioactive phenolic compounds (Saucedo-
and this represents a great advantage in many poor areas of Africa, Pompa et al., 2018). Overall, stems, leaves, flowers, pods and seeds
where its leaves are important food supplements to fight and prevent have been reported to contain phenolic acids, and in particular gallic,
malnutrition (Rodríguez-Pérez, Quirantes-Piné, Fernández-Gutiérreza, ellagic, chlorogenic and ferulic acid, together with flavonoids such as
& Segura-Carretero, 2015). Furthermore, this plant is well known for its quercetin, vanillin and kaempferol (Mbikay, 2012; Saucedo-Pompa

⁎
Corresponding authors at: Department of Pharmaceutical Sciences, Section of Food Science and Nutrition, University of Perugia, Via San Costanzo 1, Perugia,
Italy.
E-mail addresses: [email protected] (D. Montesano), [email protected] (L. Cossignani).
1
These authors contributed equally to this work and are the co-first authors.

https://doi.org/10.1016/j.foodres.2018.11.046
Received 29 August 2018; Received in revised form 19 November 2018; Accepted 22 November 2018
Available online 23 November 2018
0963-9969/ © 2018 Elsevier Ltd. All rights reserved.
G. Rocchetti et al. Food Research International 115 (2019) 319–327

et al., 2018; Singh et al., 2009; Vongsak, Sithisarn, & Gritsanapan, 2. Materials and methods
2013). Studies focusing on these compounds from Moringa have also
established important biological activities such as antioxidant and anti- 2.1. Chemicals and reagents
inflammatory properties (Falowo et al., 2018; Singh et al., 2009;
Verma, Vijayakumar, Mathela, & Rao, 2009). Gallic acid, ( ± )-6-Hydroxy-2,5,7,8-tetramethylchromane-2-car-
Generally, when plant-derived compounds are studied, the method boxylic acid (Trolox), TPTZ (2,4,6-tripyridyl-s-triazine) and AAPH
and the solvents used for extraction must be carefully considered (2,2′-azobis-2-amidinopropane dihydrochloride) were purchased from
(Vongsak et al., 2013). Phenolic compounds represent a heterogeneous Sigma-Aldrich (Milan, Italy), while fluorescein was purchased from
group of substances and they can interact with other components Invitrogen (Monza, Italy). Pure phenolic standards (purity ≥98%), in-
within the same plant matrix. Therefore, the different physical/che- cluding cyanidin, catechin, ferulic acid, resveratrol, sesamin, 5-penta-
mical nature of these compounds determines their polarity and influ- decylresorcinol, tyrosol and luteolin, were purchased from Sigma-
ences their exhaustive recovery from a plant matrix of interest (Garcia- Aldrich (St. Louis, MO, USA). Methanol and water were all LC-MS grade
Salas, Morales-Soto, Segura-Carretero, & Fernández-Gutiérrez, 2010). and purchased from VWR (Milan, Italy), while formic acid and am-
On this basis, developing different extraction methods to overcome this monium formate (both LC-MS grade) were purchased from Sigma-
issue becomes a crucial step (Garcia-Salas et al., 2010). The definition Aldrich.
of an extraction protocol must consider several aspects, among which
solvent polarity, extraction time and temperature. Consequently, total 2.2. Plant material
phenolic content and the related antioxidant activity may vary with the
variation in the extraction method (Rodríguez-Pérez et al., 2015). Fresh leaves of M. oleifera Lam. were provided by the Sud Rienergy
Generally, the knowledge of the most effective extraction conditions S.r.l. - Favella Group, a farm located in Southern Italy (Corigliano
allows to recover very important bioactives, such as flavonoids, known Calabro, Cosenza, Italy - 39°129 13′ 27,69“ N and 9° 01’ 29,69” E).
for their relevant biological properties (Fattore et al., 2016; Wang, Li, & Three samples of Moringa leaves were randomly collected in the
Bi, 2018). However, besides the extraction conditions, the actual phe- orchard from different plants from June to September 2017. The sam-
nolic content of plant materials is strongly affected by several other ples were taxonomically identified by Nicola Rizzo (Sud Rienergy
factors, such as genetic background and/or pedo-climatic conditions farm). Representative specimens of Moringa were deposited at the Orto
(Blasi, Urbani, Simonetti, Chiesi, & Cossignani, 2016; Lombardi et al., Botanico, Centro di Ateneo per i Musei Scientifici, University of Perugia
2017; Moyo, Masika, Hugo, & Muchenje, 2011). (Italy). Damaged leaves were manually discarded, while the intact ones
Several extraction techniques of bioactive compounds are reported were dried in a ventilated oven at 60 °C for 24 h (Olabode, Akanbi,
in literature, with solid-liquid extractions widely used for isolating Olunlade, & Adeola, 2015), up to constant weight. Finally, dried leaves
plant antioxidants (Castro-López et al., 2017). However, to date, the were milled in a blender and passed through 250 μm sieve to obtain a
most important classification is between conventional and non-con- thin powder having a moisture 10 ± 1%. This sample was preserved in
ventional extraction technologies (Barba et al., 2017). Conventional amber glass container, in a dry place at room temperature until pro-
extraction refers to the recovery of bioactive compounds from plant cessing.
matrices using conventional solvents (with or without heat treatment).
Plant matrix is initially homogenized and soaked in a solvent (or mix- 2.3. Extraction methods of M. oleifera leaves
ture of solvents), often under a constant agitation, thereby the desired
molecules are extracted based on diffusion and mass-transfer phe- 2.3.1. Maceration
nomena. These methods are very simple, but present drawbacks such as Three replicates (10.0 g) of dried Moringa leaves were extracted in
poor efficiency or high solvent consumption (Barba et al., 2017). On the 600 mL of both methanol 100% (MAC-1) and methanol/water 50:50 v/
other hand, non-conventional extraction techniques exploit dedicated v (MAC-2), by using a dynamic maceration, brought to room tem-
processing aids/energy inputs to improve the extraction efficiency and/ perature and kept under stirring for 4 h. Finally, the resulting extracts
or selectivity, when compared to conventional methods (Bursać were filtered, collected in an amber glass container and kept at −20 °C
Kovačević et al., 2018). Some of these methods employ ultrasounds, until the further analysis.
microwaves, supercritical fluids, and electrical/mechanical technolo-
gies in order to extract a wide range of bioactives (Formato, Gallo, 2.3.2. Homogenizer-Assisted Extraction (HAE)
Ianniello, Montesano, & Naviglio, 2013; Roselló-Soto et al., 2015). Non- Three replicates (0.5 g) of dried Moringa leaves were extracted in
conventional technologies can offer superior extraction efficiency in 30 mL of both methanol 100% (HAE-1) and methanol/water 50:50 v/v
terms of cost, yield, extraction time and/or selectivity (Barba et al., (HAE-2), by using an Ultra-turrax (Ika T25, Staufen, Germany) at
2017; Putnik et al., 2018). 5000 ×g for 3 min. The extracts were then centrifuged (Eppendorf
For these reasons, in this work, different extraction methods and 5810R, Hamburg, Germany) at 10,000 ×g for 10 min at 4 °C. Finally,
solvents have been comparatively tested to evaluate their impact on the the resulting solutions were collected in amber glass containers until
recovery of the bioactive polyphenols from Moringa leaves. An un- the further analysis.
targeted metabolomic approach, based on ultra-high-pressure liquid
chromatography (UHPLC) coupled to quadrupole-time-of-flight (QTOF) 2.3.3. Ultrasound-Assisted Extraction (UAE)
mass spectrometry, was also considered to determine the phenolic The extraction step was performed with an Ultrasonic bath mod.
composition of the extracts. In this way, a comprehensive profiling of AU-65 (ArgoLab, Carpi, Italy), composed of an inox jug with a maximal
the phenolic composition in Moringa leaves could be highlighted, in capacity of 6500 mL. Three replicates (330 mg) of Moringa leaf powder
order to shed light onto the impact of the different extraction techni- were sonicated in 20 mL of both methanol 100% (UAE-1) and me-
ques on the recovery of phenolic classes and subclasses. To the best of thanol/water 50:50 v/v (UAE-2), for 30 min at 45 °C. After the ultra-
our knowledge, this is the first attempt that deals with a comprehensive sonic treatment, the extracts were filtered and collected in amber glass
analysis of polyphenols extracted from leaves of M. oleifera cultivated in vials until the further analysis.
Italy.
2.3.4. Microwave-Assisted Extraction (MAE)
Three replicates (330 mg) of Moringa leaf samples were extracted in
20 mL of both methanol 100% (MAE-1) and methanol/water 50:50 v/v
(MAE-2). In particular, the extractions were carried out in a Biotage

320
G. Rocchetti et al. Food Research International 115 (2019) 319–327

Initiator 2.0 version 2.3, build 6250 apparatus (Biotage AB, Uppsala, (lignans), 5-pentadecylresorcinol (alkylphenols), resveratrol (stilbenes)
Sweden), setting the temperature at 45 °C and the extraction time at and tyrosol (tyrosols and other remaining phenolics) were used as re-
30 min, leaving the magnetron power controlled by the instrument in presentative of their respective classes. Results were finally expressed as
order to maintain the temperature set. After the extraction step, the mg phenolic equivalents/g dry matter (DM).
vessel used was cooled for several minutes to room temperature (25 °C),
and the obtained solution was then filtered and collected in amber glass 2.5. In vitro antioxidant activities (FRAP and ORAC)
vials until the further analysis.
The in vitro antioxidant capacity was spectrophotometrically eval-
2.3.5. Rapid Solid-Liquid Dynamic Extraction (RSLDE) uated as both ferric reducing antioxidant power (FRAP) and oxygen
Three replicates (10.0 g) of dried Moringa leaves were extracted in radical absorbance capacity (ORAC), as previously reported (Rocchetti
600 mL of both methanol 100% (NAV-1) and methanol/water 50:50 v/v et al., 2017). For both assays, a Synergy HT Multi-Detection Microplate
(NAV-2), by using a Naviglio extractor. After weighed, the leaves were Reader (BioTek Instruments Inc., Winooski, VT) was used.
included in a filter bag (Ø 50 μ) and inserted into the Naviglio extractor The FRAP working reagent consisted of 300 mM acetate buffer
chamber of Lab Model 500 cm3 capacity (Atlas Filtri Engineering, (pH 3.6), 10 mM TPTZ in 40 mM HCl and 20 mM FeCl3, considering a
Limena, Padua, Italy). The extraction conditions were the following: the ratio of 10:1:1 (v/v). The microplate was then incubated for 243 s at
program consisted in 30 cycles constituted by the alternation of a static 37 °C and the absorbance was further recorded at 600 nm. FRAP values
phase (2 min) and a dynamic phase (12 s piston down; 12 s piston up for were expressed as μmol gallic acid equivalents (GAE)/g DM.
5 cycles) and the extract was collected after 2 h. The extraction was Regarding the ORAC assay, AAPH was used as peroxyl radical
conducted at a maximum pressure of 9 bars. When the extraction generator, while fluorescein was the fluorescent probe. Wavelengths of
process was finished, the device expelled and filtered automatically the 485 nm for excitation and 528 nm for emission were used. The ORAC
extract. Besides, in order to optimize its recovery, the bag was emptied values were finally expressed as μmol Trolox equivalents (TE)/g DM.
and squeezed, and the total extract was then collected in an amber glass
container (Gallo, Conte, & Naviglio, 2017). 2.6. Statistical analysis

2.4. Profiling of polyphenols by UHPLC-QTOF mass spectrometry A one-way analysis of the variance (ANOVA) was done using the
software PASW Statistics 25.0 (SPSS Inc.) to investigate significant
An aliquot of each Moringa leaf extract was filtered using 0.22 μm differences (p < .05, Duncan's post hoc test) in FRAP, ORAC and total
cellulose syringe filters and collected in amber vials for the metabo- phenolic content (as semi-quantitative values per phenolic class) in the
lomics-based phenolic profiling. In particular, a 1290 liquid chroma- different extracts. Pearson's correlation coefficients (p = .01, two-
tograph was coupled with a G6550 mass spectrometer detector via a tailed) were also calculated in PASW Statistics 25.0.
Dual Electrospray Jet Stream ionization system (all from Agilent The Mass Profiler Professional B.12.06 (Agilent technologies) was
Technologies, Santa Clara, CA, USA). The UHPLC-QTOF instrumental formerly used for the elaboration of the untargeted UHPLC/QTOF data
conditions for the analysis of phenolic compounds in different vegetable by unsupervised hierarchical cluster analysis (HCA), based on the fold-
matrices were previously optimized (Blasi et al., 2018; Rocchetti et al., change heat map, as previously reported (Rocchetti, Lucini, et al.,
2018). Briefly, a reverse phase chromatographic separation, using a 2017). Thereafter the raw metabolomic dataset was exported and ela-
methanol-water binary gradient (5% to 90% methanol in 34 min) was borated into SIMCA 13 software (Umetrics, Malmo, Sweden) by su-
adopted. Ammonium formate (5 mM) and formic acid 0.1% (v/v) were pervised orthogonal projections to latent structures discriminant ana-
added to both phases. The mass spectrometer worked in FULL SCAN lysis (OPLS-DA) multivariate statistics. For the OPLS model, raw data
mode with a nominal resolution at 30,000 FWHM and in positive po- were Log2 transformed, UV scaled and then analyzed by OPLS-DA. The
larity. For the mass-acquisition, a range of 100–1200 m/z was set up. variation between the groups was separated into predictive and or-
The injection volume was 3.5 μL considering three replicates for each thogonal (i.e., related to technical and biological variation) compo-
sample. nents. The presence of outliers into the OPLS model was checked ac-
Raw data were aligned and deconvoluted using the Agilent cording to Hotelling's T2 (i.e., the distance from the origin in the
Profinder B.06 software and then annotated according to the ‘find-by- model), using 95% and 99% confidence limits for suspect and strong
formula’ algorithm, using the combination of monoisotopic accurate outliers, respectively. The model cross-validation was then carried out
mass and the entire isotopic pattern (i.e., isotopic spacing and isotopic using CV-ANOVA (p < .01), whereas permutation testing (N = 100)
ratio) against Phenol-Explorer 3.6 (Rothwell et al., 2013), the com- was done to exclude overfitting. Model parameters, i.e., R2Y (goodness-
prehensive database on phenolic compounds available online. A mass of-fit) and Q2Y (goodness-of-prediction) were also recorded.
accuracy tolerance below 5 ppm was used for annotation purposes, and Afterwards, the variables selection methods namely VIP (i.e., vari-
features were recursively identified using a retention time tolerance able importance in projection) was used to evaluate the importance of
of ± 0.1 min for alignment (Lucini et al., 2016). This approach allowed each phenolic compounds in discriminating the different extraction
gaining a higher confidence in the annotation step and was in com- techniques, and to select those having the highest discrimination po-
pliance to the Level 2 of identification (i.e., putatively annotated com- tential (VIP score > 1.2). Finally, a Fold-Change (FC) analysis was
pounds), as previously reported by COSMOS Metabolomics Standards carried out considering the UHPLC/QTOF data, in order to evaluate in
Initiative (http://cosmos-fp7.eu/msi). The Agilent Profinder B.06 soft- detail, the impact of each extraction method on the different phenolic
ware was used also for the post-acquisition data filtering: only those subclasses.
compounds putatively annotated within 100% of replications in at least
one condition were retained. 3. Results and discussion
To achieve a data reduction from the complexity of compounds
annotated, as well as to provide more quantitative information, phe- 3.1. Phenolic profiling and antioxidant activity of Moringa leaves through
nolic compounds were firstly ascribed into classes and subclasses, and different extraction methods
then quantified using methanolic standard solutions (methanol/water
80:20, v/v) of single phenolics analyzed through the same method In this work, a comprehensive UHPLC-ESI-QTOF mass spectrometry
(Rocchetti et al., 2017). Cyanidin (anthocyanins), catechin (flavonols approach was used to investigate in an unbiased manner the phenolic
and flavanols), luteolin (flavones and other remaining flavonoids), composition of Moringa leaf samples subjected to different extraction
ferulic acid (hydroxycinnamic acids and other phenolic acids), sesamin methods. Overall, 262 compounds were putatively annotated when

321
G. Rocchetti et al. Food Research International 115 (2019) 319–327

Table 1 conditions, both UAE-1 and UAE-2 extractions were more performing
Total phenolic content (as gained by UHPLC-ESI-QTOF mass spectrometry) and than MAC-1 and MAC-2, thus corroborating their results. However,
in vitro antioxidant activity (as both FRAP reducing power and ORAC radical these authors used a spectrophotometric assay (i.e. the Folin-Ciocalteu)
scavenging). Data (expressed on a dry matter basis, DM) are presented as mean to assess the TPC content of their extract, a method that has been dis-
values ± standard deviation (n = 3).
couraged because presenting many limitations and pitfalls (Granato
Sample Extraction type Total phenolics FRAP (μmol ORAC (μmol et al., 2018). Indeed, high-resolution methods based on UHPLC coupled
(mg Eq./g DM) GAE/g DM) TE/g DM) to mass spectrometry are strongly advisable to better characterize the
phenolic composition of different plant-extract (Granato et al., 2018).
NAV-1 RSLDE 7.59 ± 0.56a 1.19 ± 0.09a 11.77 ± 2.14a
(MeOH 100%) When specifically referring to each phenolic class characterizing Mor-
NAV-2 RSLDE 15.35 ± 1.47c 1.04 ± 0.04a 13.37 ± 0.08a inga leaf extracts (Fig. 1), it was evident that phenolic acids and fla-
(MeOH:H2O vonol equivalents were those groups of compounds possessing the
50:50, v/v)
widest distribution, regardless of the extraction method considered. In
MAE-1 Microwave 22.40 ± 2.55e 20.57 ± 1.12c 252.01 ± 6.47d
(MeOH 100%)
this regard, the average content of phenolic acids detected was
MAE-2 Microwave 16.97 ± 0.85cd 35.24 ± 2.14e 59.77 ± 6.55b 7.17 mg/g DM, with the highest values recorded in HAE-1 and UAE-1
(MeOH:H2O extracts (i.e., 14.22 and 12.37 mg/g DM, respectively). The average
50:50, v/v) content of flavonols detected was 2.95 mg/g DM, however some ex-
MAC-1 Maceration 10.27 ± 1.26b 9.01 ± 0.18b 114.52 ± 5.61c
tracts, such as MAC-1, MAC-2 and NAV-1, showed very low content
(MeOH 100%)
MAC-2 Maceration 5.62 ± 0.07a 3.28 ± 0.61a 49.77 ± 1.38b when compared to the others. Furthermore, tyrosol and alkylphenol
(MeOH:H2O equivalents were found to be very abundant, with average values of
50:50, v/v) 2.57 and 1.77 mg/g DM, respectively. Interestingly, lignans and an-
UAE-1 Ultrasound 26.36 ± 0.40f 19.27 ± 1.88c 292.24 ± 6.69e thocyanins were annotated in each extract analyzed, recording the
(MeOH 100%)
UAE-2 Ultrasound 14.48 ± 0.40c 36.52 ± 2.59e 324.54 ± 7.73g
highest values in HAE-1 extracts (i.e., 2.06 and 1.66 mg/g DM, re-
(MeOH:H2O spectively).
50:50, v/v) To date, different studies have reported the presence of bioactive
HAE-1 Homogenization 35.19 ± 2.14g 25.82 ± 2.22d 536.27 ± 1.97h phenolic compounds from Moringa tissues, describing the corre-
(MeOH 100%)
sponding in vitro antioxidant activity (Saucedo-Pompa et al., 2018).
HAE-2 Homogenization 18.75 ± 1.75d 35.78 ± 0.65e 305.33 ± 9.22f
(MeOH:H2O Therefore, most of the research has been focused on optimizing the best
50:50, v/v) extractive conditions of these compounds from the Moringa plant and/
or derived products (Saucedo-Pompa et al., 2018). When considering
Superscript letters within each column indicate homogeneous sub-class as re- M. oleifera leaves, some authors reported that flavonoids were the
sulted from ANOVA (p < .01), Duncan's post-hoc test. predominant group of phenolics in this matrix (Makita, Chimuka,
Steenkamp, Cukrowska, & Madala, 2016; Rodríguez-Pérez et al., 2015),
considering all the samples analyzed, with a clear abundance of fla- with a clear abundance of glycosylated forms of quercetin, kaempferol
vonoids (117 compounds) compared to the other phenolic classes. In and isorhamnetin derivatives (Coppin et al., 2013). Furthermore, it has
this regard, flavone equivalents were the most frequent compounds (46 been also reported that hydroxycinnamics, such as caffeoylquinic and
annotations) followed by anthocyanins and flavonols (36 and 35 com- coumaroylquinic isomers, were the main phenolic acids characterizing
pounds annotated, respectively). The second most frequent class was Moringa plants, although geographic and pedo-climatic conditions
represented by phenolic acids (60 compounds), among which hydro- could affect their final composition (Nouman et al., 2016; Saucedo-
xycinnamic acids were the most common (47 annotations) followed by Pompa et al., 2018). In our experimental conditions, we found that
a lower number of hydroxybenzoics, hydroxyphenylacetics and hy- glycosidic and isomeric forms of quercetin (i.e., quercetin 3-O-galacto-
droxyphenylpropionics. Interestingly, other phenolic classes such as side, quercetin 3-O-glucoside and quercetin 3-O-sophoroside), kaemp-
lignans, stilbenes and lower-molecular-weight phenolics such as tyrosol ferol (i.e., kaempferol 3-O-galacoside, kaempferol 3-O-glucoside and
equivalents could be annotated within the dataset, thus suggesting a kaempferol 3-O-rutinoside) and isorhamnetin (i.e., isorhamnetin 3-O-
wide chemical diversity across phenolics. A detailed list of all poly- galactoside/glucoside together with isorhamnetin 7-O-rhamnoside)
phenols annotated in each Moringa leaf sample according to the dif- were the most abundant flavonols, thus corroborating the outcomes
ferent extraction methods is provided as supplementary material, to- reported in literature. Different results were obtained when considering
gether with composite mass spectra. the phenolic acids distribution (supplementary material); in fact, al-
Afterwards, by using the curves from pure phenolic standard com- though isomeric forms of caffeoylquinic and coumaroylquinic acids
pounds, the semi-quantitative values for each phenolic class detected could be annotated, the most abundant compounds were isomeric forms
were calculated and the cumulative contents are provided in Table 1 of hydroxybenzoic acid (i.e., 2/3/4-hydroxybenzoic acids) together
and summarized in Fig. 1, as mg phenolic equivalents/g DM. In parti- with cinnamoyl glucose (belonging to the hydroxycinnamic acids). The
cular, HAE-1 extracts showed the highest (p < .05) total phenolic lower-molecular-weight phenolics such as protocatechuic aldehyde and
content (TPC), being 35.19 mg/g, followed by UAE-1 and MAE-1 ex- phlorin, together with the lignan sesamol could be found in high con-
tracts (26.36 and 22.40 mg/g DM, respectively). Besides, the lowest centrations in all the samples analyzed (supplementary material). In-
(p < .05) TPCs were recorded in MAC-2 (5.62 mg/g DM) and NAV-1 terestingly, this can be considered one of the first works identifying, by
(7.59 mg/g DM) extracts. Overall, Moringa leaves extracted in pure means of high-resolution mass spectrometry, glycosidic forms of cya-
methanol showed higher phenolic contents when compared with the nidin and delphinidin as the main anthocyanins characterizing M.
hydroalcoholic (methanol/water 50:50, v/v) extracts. This was true for oleifera leaves.
all samples except for those leaves subjected to a RSLDE extraction (i.e., The interest in phenolic compounds characterizing Moringa leaves
NAV-1 and NAV-2 extracts), showing in fact an opposite trend. In a is due also to their bioactive properties, such as the antioxidant po-
previous work, Rodríguez-Pérez et al. (2015) tested a conventional tential. Therefore, in this work, the in vitro antioxidant activity of
solid-liquid extraction (maceration) and UAE for optimizing the re- Moringa leaves was investigated by means of two complementary tests,
covery of phenolic compounds from M. oleifera leaves. These authors namely FRAP and ORAC. The aforementioned methods are based on
reported that UAE using ethanol/water (50:50, v/v) was the best ex- two different reaction kinetics, being a single electron transfer and a
traction procedure, allowing to recover 47 mg GAE/g dry leaf hydrogen atom transfer, respectively, and then can be considered quite
(Rodríguez-Pérez et al., 2015). Indeed, also in our experimental representative of the mechanisms of action of antioxidant compounds

322
G. Rocchetti et al. Food Research International 115 (2019) 319–327

Fig. 1. Cumulative phenolic composition (as mg phenolic equivalents/g dry matter, DM) characterizing the different Moringa leaf extracts.

(Alam, Bristi, & Rafiquzzaman, 2013). The individual FRAP and ORAC antioxidant activity from Moringa leaves.
values from Moringa leaf samples subjected to different extraction
methods/conditions are provided in Table 1. Overall, the in vitro anti-
3.2. Comparison of different extraction methods on recovery of phenolic
oxidant activities were significantly affected by the extraction methods
subclasses
used. In particular, the FRAP values (as μmol GAE/g DM) ranged from
1.04 in NAV-2 extracts up to 36.52 in UAE-2 extracts; in this regard, the
In order to compare the different extraction technologies tested,
highest values (p < .05) were recorded in UAE-2 extracts followed by
multivariate statistics were applied to the metabolomic dataset.
HAE-2 and MAE-2 extracts (being 35.78 and 35.24 μmol GAE/g DM,
Multivariate analysis of metabolomics-based data is usually performed
respectively). Different results were obtained looking at the ORAC va-
by using unsupervised (such as hierarchical cluster analysis, HCA) and
lues, recording the highest activity for HAE-1 extracts (536.27 μmol TE/
supervised statistical tools (such as OPLS discriminant analysis).
g DM) and the lowest in NAV-1 (11.77 μmol TE/g DM). Looking to si-
Therefore, the unsupervised HCA was firstly produced from the
milar works in literature, Castro-López et al. (2017) found lower FRAP
fold-change heat map in order to group and visualize the different ex-
reducing power in maceration extracts when compared with those ob-
tracts according to intrinsic similarities in their measurements, irre-
tained by using MAE and UAE technologies, thus confirming our re-
spective of samples class memberships. As showed in Fig. 2, two main
sults. Furthermore, a possible explanation to the high FRAP values re-
clusters were obtained when considering the different Moringa leave
corded in MAE extracts could be the generation of new reducing
extracts, with all biological replicates of each treatment clustering to-
compounds during MAE process, mainly from sugars characterizing the
gether. The first cluster was characterized by MAC-1 and NAV-1 ex-
plant matrix, as reported by Charurin, Ames, and Del Castillo (2002).
tracts, while all the other samples were included in the second cluster.
Regarding ORAC assay, Castillo-López et al. (2017) showed an average
Remarkably, the HCA suggested that the extraction type was hier-
ORAC value of 168.5 μmol TE/g DM, analysing methanolic extracts of
archically more important than the extraction solvent. In fact, MAC-2
Moringa leaves from Mexico obtained with HAE. Therefore, our HAE
extracts were close to methanolic extracts into the heat map, thus
methanolic extracts were found to be characterized by 3-time higher
suggesting a clear effect related to the extraction type performed. An-
values (i.e., 536.27 vs 168.5).
other interesting information was related to MAC-1 and NAV-1 extracts;
Afterwards, Pearson's correlation coefficients (supplementary ma-
in particular, it was evident that a sub-cluster of compounds was clearly
terial) were calculated in order to shed light on the contribution from
absent in those extracts when compared to the others (Fig. 2). Gen-
the different phenolic classes to the in vitro antioxidant activities ob-
erally, the use of class membership criteria characterizing supervised
served. Interestingly, only flavonols, flavones and anthocyanins sig-
methods allows a better separation between classes into the score plot
nificantly correlated with FRAP values (i.e., 0.62, 0.55 and 0.48, re-
space. Furthermore, the OPLS-DA approach is also able to effectively
spectively; p < .01), while completely different results were obtained
separate Y-predictive variation from Y-uncorrelated variation in X (data
when considering ORAC. In fact, all the phenolic classes considered
matrix). Therefore, an OPLS-DA score plot was built considering each
showed high and significant correlation coefficients with the ORAC
different extraction condition, as provided in Fig. 3. In our experimental
values recorded, with phenolic acids possessing the highest value (0.77;
conditions, a complete separation between methanolic and hydroalco-
p < .01), followed by tyrosols and anthocyanins (0.72; p < .01).
holic extracts was achieved, with clear differences related above all to
These trends were confirmed by correlating the TPC (as assessed by
the different extraction technologies. Notably, the discriminant model
UHPLC-QTOF semi-quantitative analysis) with both FRAP and ORAC
indicators were found to be excellent, being R2Y (the goodness-of-
values. In fact, a higher coefficient was recorded between TPC and
fit) = 0.98 and Q2Y (goodness-of-prediction) = 0.88, with more than
ORAC (0.82; p < .01) when compared with TPC and FRAP (0.47;
adequate cross-validation parameters (supplementary material), thus
p < .01). Finally, a significant correlation coefficient was found also
confirming the robustness of the OPLS model based on the phenolic
between FRAP and ORAC activities (0.62; p < .01). The above results
profiles for discriminating the different extraction methods. Overall, the
indicated that extraction technologies such as HAE, UAE and MAE
OPLS-DA confirmed the findings outlined by the HCA. In particular,
could be promising tools for extracting bioactive compounds possessing
looking at methanolic extracts, the first latent vector of the OPLS score

323
G. Rocchetti et al. Food Research International 115 (2019) 319–327

Fig. 2. Non-averaged unsupervised cluster analysis on the phenolic profile of different Moringa leaf extracts (similarity: ‘Euclidean’; linkage rule: ‘Ward’).
Compounds intensity was used to build up heat map, on the basis of which the clusters were generated.

Fig. 3. Orthogonal projection to latent structures discriminant analysis (OPLS-DA) on different Moringa leaf extracts phenolic profiles. Individual replications are
given in the class prediction model score plot.

324
G. Rocchetti et al. Food Research International 115 (2019) 319–327

Table 2
Discriminant phenolic compounds identified by VIP (variable importance in projection) selection method following supervised OPLS-DA modelling. Compounds are
provided together with VIP scores (measure of variables importance in the OPLS model).
Phenolic class Phenolic subclass Marker VIP score

Flavonoids Anthocyanins Pinotin A 1.46 ± 0.56

Petunidin 3-O-(6″-p-coumaroyl-glucoside) 1.46 ± 0.56
Pelargonidin 3-O-galactoside 1.33 ± 0.51
Pelargonidin 3-O-glucoside 1.33 ± 0.51
Delphinidin 3-O-(6″-acetyl-glucoside) 1.32 ± 0.45
Delphinidin 3-O-(6”acetyl-galactoside) 1.32 ± 0.45
Dihydrochalcones 3-Hydroxyphloretin 2’-O-glucoside 1.27 ± 0.73
Dihydroflavonols Dihydroquercetin 3-O-rhamnoside 1.25 ± 0.47
Dihydroquercetin 1.20 ± 0.49
Flavanols Catechin 3-O-glucoside 1.27 ± 0.73
Flavanones Naringenin 7-O-glucoside 1.27 ± 0.73
Naringin 4’-O-glucoside 1.27 ± 0.73
Eriodictyol 7-O-glucoside 1.25 ± 0.46
Flavones Hispidulin 1.46 ± 0.48
Jaceosidin 1.28 ± 0.71
Apigenin 6-C-glucoside 1.26 ± 0.46
Flavonols Kaempferol 3-O-rhamnoside 1.49 ± 1.15
3,7-Dimethylquercetin 1.28 ± 0.71
Isoflavonoids Genistin 1.26 ± 0.46
Phenolic acids Hydroxybenzoics Vanillic acid 1.80 ± 0.39
Syringic acid 1.59 ± 0.81
Gallic acid ethyl ester 1.59 ± 0.81
Hydroxycinnamics 1-Sinapoyl-2,2′-diferuloylgentiobiose 1.68 ± 0.71
Schottenol/Sitosterol ferulate 1.66 ± 0.83
24-Methylcholestanol ferulate 1.39 ± 0.70
Ferulic acid 4-O-glucoside 1.33 ± 0.57
Feruloyl glucose 1.33 ± 0.57
1,2,2′-Triferuloylgentiobiose 1.32 ± 0.83
3/4/5-Sinapoylquinic acid 1.30 ± 0.62
Avenanthramide 2f 1.28 ± 0.71
Ferulic/Isoferulic acid 1.24 ± 0.42
Hydroxyphenylacetics 3,4-Dihydroxyphenylacetic acid 1.79 ± 0.38
Other polyphenols Alkylphenols 5-Heptadecylresorcinol 1.55 ± 0.15
5-Pentacosylresorcinol 1.49 ± 0.98
5-Nonadecylresorcinol 1.32 ± 0.84
5-Heneicosylresorcinol 1.30 ± 1.07
5-Pentacosenylresorcinol 1.20 ± 1.15
Alkylmethoxyphenols 4-Vinylsyringol 1.33 ± 0.69
Hydroxycoumarins Umbelliferone 1.32 ± 1.11
4-Hydroxycoumarin 1.32 ± 1.11
Hydroxyphenylpropenes Estragole 1.30 ± 0.63
[6]-Gingerol 1.21 ± 0.75
Others Pyrogallol 1.27 ± 0.34
Phenolic terpenes Rosmanol/Epirosmanol 1.32 ± 0.54
Tyrosols Hydroxytyrosol 4-O-glucoside 1.28 ± 0.73

plot clearly allowed discriminating MAC-1 and NAV-1 from the other followed by glycosidic forms of pelargonidin, delphinidin and petunidin
extracts, thus confirming what observed into the hierarchical clus- (anthocyanins). However, the highest VIP scores (> 1.5) were recorded
tering. Notably, UAE-1 possessed a different phenolic profile when for 7 phenolic acids, namely vanillic acid, 3,4-dihydroxyphenylacetic
compared with MAE-1 and HAE-1 extracts. Regarding the hydroalco- acid, schottenol and sitosterol ferulate, syringic acid, gallic acid ethyl
holic extracts, all the extraction technologies provided similar profiles, ester, 1-sinapoyl-2,2′-diferuloylgentiobiose, and one alkylphenol (5-
with the exception of MAC-2 extracts, again consistently with HCA. It is heptadecylresorcinol). Therefore, this statistical approach indicates that
also important to notice that the OPLS-DA score plot was coherent with phenolic acids are those compounds most affected by the different ex-
the TPC outcomes (Fig. 1); in fact, MAC-1, MAC-2 and NAV-1, i.e. those traction technologies tested, followed by the most represented class of
extracts characterized by a lower TPC, were also those possessing the phenolics such as flavonoids.
most differential profile (Fig. 3). Subsequently, the variables' im- Afterwards, a Fold-Change (FC) analysis was used to evaluate the
portance in the OPLS-DA model was evaluated by means of VIP (i.e., recovery of each phenolic subclass identified by UHPLC-QTOF profiling
variable importance in projections) statistical tool. This approach is (supplementary material). This goal was achieved by comparing the
able to summarize those compounds possessing the highest dis- phenolic subclasses characterizing each extract against HAE-1 extract,
crimination potential into the predictive model. The phenolic com- i.e., that extract showing the highest TPC (Fig. 1 and Table 1). Inter-
pounds characterized by a VIP score > 1.2 were recorded and finally estingly, MAE-1 was found to be the most effective method to recover
reported in Table 2, classified into phenolic classes and subclasses. eight phenolic subclasses, namely anthocyanins (FC = 1.09), dihydro-
These compounds are likely those better summarizing the differences in flavonols (FC = 1.22), flavones (FC = 1.25), flavonols (FC = 1.13),
extraction efficiencies across the different protocols tested. As a general isoflavonoids (FC = 1.46), alkylmethoxyphenols (FC = 1.39), fur-
consideration, 50 compounds possessed the score threshold, with an anocoumarins (FC = 1.96) and hydroxybenzoketones (FC = 2.63).
abundance of flavonoids (19 compounds), followed by phenolic acids NAV-1 was the only extract recording chalcones; nonetheless, this was
(17 compounds) and other lower-molecular-weight compounds. The also the extraction technology allowing to recover the highest values of
hydroxycinnamic acids were the most abundant subclass of compounds, stilbenes (FC = 1.79), hydroxycoumarins (FC = 1.29) and other

325
G. Rocchetti et al. Food Research International 115 (2019) 319–327

phenolics (FC = 1.25, including phlorin). Curcuminoids (such as bis- acids being the most represented classes of compounds). The following
demethoxycurcumin and demethoxycurcumin) were better extracted use of both unsupervised and supervised multivariate statistics applied
by means of HAE-2 (FC = 2.25), followed by NAV-2 (FC = 1.88), UAE- to the metabolomic dataset allowed also to investigate the classes of
2 (FC = 1.77) and MAE-2 (FC = 1.70), thus suggesting a clear influence compounds being mostly affected by the different extractions tested. In
of the hydroalcoholic solution on the recovery of these compounds. this regard, the recovery of phenolic acids was the parameter being the
Finally, hydroxycinnamaldehydes (i.e., ferualdehyde and sinapalde- most affected by the different extraction technologies. After studying
hyde) characterized exclusively UAE-1 extracts, recording a FC value of ten different extraction procedures (using different extraction ap-
1.14. proaches and either methanol 100% or methanol/water 50:50 v/v as
In this work, we tested four non-conventional extraction methods extraction solvents) we found that HAE using methanol 100% produced
(MAE, HAE, UAE and RSLDE) for recovering antioxidant phenolic an extract of M. oleifera leaves with the highest amounts of phenolic
compounds from M. oleifera leaves. In addition, maceration was used as compounds. Besides, the highest FRAP values were recorded for UAE-2,
conventional technique. Generally, the major drawbacks of conven- HAE-2 and MAE-2 samples (being 36.52, 35.78 and 35.24 μmol GAE/g,
tional methods are longer extraction times, requirement of the right and respectively), while the highest ORAC value (i.e., 536.27 μmol TE/g)
high amount of extraction solvent (in high purity), evaporation of huge was found for the HAE-1 sample. By exploiting the comprehensive
amounts of solvent, low extraction selectivity and the possible de- metabolomics-based approach, we found that each extract showed ex-
gradation of thermolabile antioxidant compounds due to the high clusively subclasses of polyphenols, thus underlying the importance of
working temperatures used (Selvamuthukumaran & Shi, 2017). Overall, choosing the right extraction method for a correct and wider com-
non-conventional extraction techniques are trying to overcome these pounds characterization. Finally, this work suggests the use of non-
latter limitations. In our experimental conditions, the maceration took conventional extraction technologies for extracting the polyphenols
place at room temperature, and therefore it is expected that there was potentially available from plant materials.
no alteration of the thermolabile compounds. Despite observing dif-
ferences among the different phenolics subclasses, we found that HAE Conflict of interest
(using methanol 100% as extraction solvent) was the most efficient
technique in recovering the highest phenolic content. This extraction The authors declare no conflict of interest.
method promotes the rupture of the plant cell structures by exploiting Supplementary data to this article can be found online at https://
the high shear rate within a few seconds, thus allowing the release of doi.org/10.1016/j.foodres.2018.11.046.
metabolites such as polyphenols. Furthermore, this technique requires a
lower consumption of solvent and time, together with low energy re- Acknowledgements
quirements when compared with the other non-conventional methods
(Pereira, Molina, Arruda, & Pastore, 2017). GR and SG were the recipient of a Ph.D. fellowship from the
In a recent work, Castro-López et al. (2017) demonstrated that the Università Cattolica del Sacro Cuore (UCSC, Piacenza, Italy). This work
extraction method had a strong influence on the recovery of poly- was supported by funding from University of Perugia: Fondo d'Ateneo
phenols from M. oleifera leaves. In particular, they found that MAE and per la Ricerca di Base 2015, Project “Caratterizzazione chimico-anali-
UAE allowed to obtain the highest recovery of phenolic compounds, tica e nutrizionale di componenti bioattivi in prodotti vegetali e po-
followed by MAC, thus corroborating our findings. During MAE, the tenziale utilizzo nella formulazione di alimenti funzionali”. The authors
combination of temperature/microwave energy is accounted to burst wish to thank the “Romeo and Enrica Invernizzi” foundation for sup-
the cell wall, thus releasing polyphenols into the extraction solvent porting the metabolomic platform and the Sud Rienergy Soc. Agr. S.r.l. -
more effectively (Kaufmann & Christen, 2002). Instead, UAE is a pro- Favella Group (Corigliano Calabro, CS, Italy) for providing the bota-
cess promoting the extraction of the plant bioactives by exploiting the nical material.
mechanical action of ultrasound on cell structures, thus decreasing the
times of transfer of these substances from the plant material to ex- References
traction solvent. One of the major drawbacks is related to the scarce
possibility to carry out a selective extraction, regardless of the affinity Alam, M. N., Bristi, N. J., & Rafiquzzaman, M. (2013). Review on in vivo and in vitro
with the solvent used (Gallo et al., 2017). Furthermore, in this work, we methods evaluation of antioxidant activity. Saudi Pharmaceutical Journal, 21,
143–152.
used a recent “green” non-conventional technology based on rapid Anwar, F., Ashraf, M., & Gilani, A. H. (2007). Moringa oleifera: A food plant with multiples
solid-liquid dynamic extraction. This technology exploits the Naviglio medicinal uses. Phytotherapy Research, 21, 17–25.
extractor and is based on the use of a negative pressure gradient from Barba, F. J., Putnik, P., Kovacevic, D. B., Poojary, M. M., Roohinejad, S., Lorenzo, J. M., &
Koubaa, M. (2017). Impact of conventional and non-conventional processing on
the inside to outside of the solid matrix. This strategy was recently prickly pear (Opuntia spp.) and their derived products: From preservation of bev-
tested for both reproducibility and recovery of bioactive components erages to valorization of by-products. Trends in Food Science and Technology, 67,
from different plant materials (Formato et al., 2013; Gallo et al., 2017; 260–270.
Blasi, F., Rocchetti, G., Montesano, D., Lucini, L., Chiodelli, G., Ghisoni, S., ... Cossignani,
Naviglio, Montesano, & Gallo, 2015), showing promising results.
L. (2018). Changes in extra-virgin olive oil added with Lycium barbarum L. car-
Looking to our findings, it can be postulated that the use of a hydro- otenoids during frying: Chemical analyses and metabolomic approach. Food Research
alcoholic solvent promoted the best recovery of phenolic compounds International, 105, 507–516.
Blasi, F., Urbani, E., Simonetti, M. S., Chiesi, C., & Cossignani, L. (2016). Seasonal var-
when the Naviglio extractor was used (Table 1 and Fig. 1). Overall, we
iations in antioxidant compounds of Olea europaea leaves collected from different
can state that non-conventional technologies are promising strategies Italian cultivars. Journal of Applied Botany and Food Quality, 89, 202–207.
for enhancing the recovery of bioactive and antioxidant polyphenols Bursać Kovačević, D., Maras, M., Barba, F. J., Granato, D., Roohinejad, S., Mallikarjunan,
from Moringa leaves. K., ... Putnik, P. (2018). Innovative technologies for the recovery of phytochemicals
from Stevia rebaudiana Bertoni leaves: A review. Food Chemistry, 268, 513–521.
Castillo-López, R. I., Léon-Félix, J., Angulo-Escalante, M. A., Gutíerrez-Dorado, R., Muy-
4. Conclusions Rangel, M. D., & Heredia, J. B. (2017). Nutritional and phenolic characterization of
Moringa oleifera leaves grown in Sinaloa. México. Pakistan Journal of Botany, 49,
161–168.
In this work, the impact of different extraction technologies (in- Castro-López, C., Ventura-Sobrevilla, J. M., González-Hernández, M. D., Rojas, R.,
cluding conventional and non-conventional), namely RSLDE, MAE, Ascacio-Valdés, J. A., Aguilar, C. N., & Martínez-Ávila, G. C. C. (2017). Impact of
MAC, HAE and UAE, was evaluated in terms of recovering the phenolic extraction techniques on antioxidant capacities and phytochemical composition of
polyphenol-rich extracts. Food Chemistry, 237, 1139–1148.
profile of M. oleifera leaves. Phenolic profiles were determined by using Charurin, P., Ames, J. M., & Del Castillo, M. D. (2002). Antioxidant activity of coffee
an untargeted UHPLC/QTOF mass spectrometric approach, allowing to model systems. Journal of Agricultural and Food Chemistry, 50, 3751–3756.
putatively annotate > 260 polyphenols (with flavonoids and phenolic Coppin, J. P., Xu, Y., Chen, H., Pan, M.-H., Ho, C.-T., Juliani, R., ... Wu, Q. (2013).

326
G. Rocchetti et al. Food Research International 115 (2019) 319–327

Determination of flavonoids by LC/MS and anti-inflammatory activity in Moringa homogenizer-assisted extraction (HAE) of total phenolic compounds from banana
oleifera. Journal of Functional Foods, 5, 1892–1899. peel. Journal of Food Process Engineering, 40, e12438.
Falowo, A. B., Mukumbo, F. E., Idamokoro, E. M., Lorenzo, J. M., Afolayan, A., & Putnik, P., Lorenzo, J. M., Barba, F. J., Roohinejad, S., Jambrak, A. R., Granato, D., ...
Muchenje, V. (2018). Multi-functional application of Moringa oleifera Lam. in nutri- Bursać Kovačević, D. (2018). Novel food processing and extraction technologies of
tion and animal food products: A review. Food Research International, 106, 317–334. high-added value compounds from plant materials. Food, 7, 106–122.
Fattore, M., Montesano, D., Pagano, E., Teta, R., Borrelli, F., Mangoni, A., ... Albrizio, S. Rocchetti, G., Chiodelli, G., Giuberti, G., Ghisoni, S., Baccolo, G., Blasi, F., ... Lucini, L.
(2016). Carotenoid and flavonoid profile and antioxidant activity in “Pomodorino (2018). UHPLC-ESI-QTOF-MS profile of polyphenols in Goji berries (Lycium barbarum
Vesuviano” tomatoes. Journal of Food Composition and Analysis, 53, 61–68. L.) and its dynamics during in vitro gastrointestinal digestion and fermentation.
Formato, A., Gallo, M., Ianniello, D., Montesano, D., & Naviglio, D. (2013). Supercritical Journal of Functional Foods, 40, 564–572.
fluid extraction of α- and β-acids from hops compared to cyclically pressurized solid- Rocchetti, G., Chiodelli, G., Giuberti, G., Masoero, F., Trevisan, M., & Lucini, L. (2017).
liquid extraction. Journal of Supercritical Fluids, 84, 113–120. Evaluation of phenolic profile and antioxidant capacity in gluten-free flours. Food
Gallo, M., Conte, E., & Naviglio, D. (2017). Analysis and comparison of the antioxidant Chemistry, 228, 367–373.
component of Portulaca oleracea leaves obtained by different solid-liquid extraction Rocchetti, G., Lucini, L., Chiodelli, G., Giuberti, G., Montesano, D., Masoero, F., &
techniques. Antioxidants, 6, 64–73. Trevisan, M. (2017). Impact of boiling on free and bound phenolic profile and anti-
Garcia-Salas, P., Morales-Soto, A., Segura-Carretero, A., & Fernández-Gutiérrez, A. oxidant activity of commercial gluten-free pasta. Food Research International, 100,
(2010). Phenolic-compound-extraction systems for fruit and vegetable samples. 69–77.
Molecules, 15, 8813–8826. Rodríguez-Pérez, C., Quirantes-Piné, R., Fernández-Gutiérreza, A., & Segura-Carretero, A.
Granato, D., Shahidi, F., Wrolstad, R., Kilmartin, P., Melton, L. D., Hidalgo, F. J., ... (2015). Optimization of extraction method to obtain a phenolic compounds-rich in
Finglas, P. (2018). Antioxidant activity, total phenolics and flavonoids contents: native and callus culture of Moringa oleifera Lam. Journal of Medical and
Should we ban in vitro screening methods? Food Chemistry, 264, 471–475. Bioengineering, 2, 163–167.
Kaufmann, B., & Christen, P. (2002). Recent extraction techniques for natural products: Roselló-Soto, E., Galanakis, C. M., Brnčić, M., Orlien, V., Trujillo, F. J., Mawson, R., ...
Microwave-assisted extraction and pressurized solvent extraction. Phytochemical Barba, F. (2015). Clean recovery of antioxidant compounds from plant foods, by-
Analysis, 13, 105–113. products and algae assisted by ultrasounds processing. Modeling approaches to op-
Lombardi, G., Cossignani, L., Giua, L., Simonetti, M. S., Maurizi, A., Burini, G., ... Blasi, F. timize processing conditions. Trends in Food Science & Technology, 42, 134–149.
(2017). Phenol composition and antioxidant capacity of red wines produced in Rothwell, J. A., Pérez-Jiménez, J., Neveu, V., Medina-Ramon, A., M'Hiri, N., Garcia
Central Italy changes after one-year storage. Journal of Applied Botany and Food Lobato, P., ... Scalbert, A. (2013). Phenol-Explorer 3.0: A major update of the Phenol-
Quality, 90, 197–204. Explorer data-base to incorporate data on the effects of food processing on poly-
Lucini, L., Borgognone, D., Rouphael, Y., Cardarelli, M., Bernardi, J., & Colla, G. (2016). phenol content. Database: The Journal of Biological Databases and Curation. https://
Mild potassium chloride stress alters the mineral composition, hormone network, and doi.org/10.1093/database/bat070.
phenolic profile in Artichoke leaves. Frontiers in Plant Science, 7, 948. Saucedo-Pompa, S., Torres-Castillo, J. A., Castro-López, C., Rojas, R., Sánchez-Alejo, E. J.,
Makita, C., Chimuka, L., Steenkamp, P., Cukrowska, E., & Madala, E. (2016). Comparative Ngangyo-Heya, M., & Martínez-Ávilla, G. C. G. (2018). Moringa plants: Bioactive
analyses of flavonoid content in Moringa oleifera and Moringa ovalifolia with the aid of compounds and promising applications in food products. Food Research International,
UHPLC-qTOF-MS fingerprinting. South African Journal of Botany, 105, 116–122. 111, 438–450.
Mbikay, M. (2012). Therapeutic potential of Moringa oleifera leaves in chronic hy- Selvamuthukumaran, M., & Shi, J. (2017). Recent advances in extraction of antioxidants
perglycemia and dyslipidemia: A review. Frontiers in Pharmacology, 3, 24. from plant by-products processing industries. Food Quality and Safety, 1, 61–81.
Moyo, B., Masika, P. J., Hugo, A., & Muchenje, V. (2011). Nutritional characterization of Singh, B. N., Singh, B. R., Singh, R. L., Prakash, D., Dhakarey, R., Upadhyay, G., & Singh,
Moringa (Moringa oleifera Lam.) leaves. African Journal of Biotechnology, 10, H. B. (2009). Oxidative DNA damage protective activity, antioxidant and anti-
12925–12933. quorum sensing potentials of Moringa oleifera. Food and Chemical Toxicology, 47,
Naviglio, D., Montesano, D., & Gallo, M. (2015). Laboratory production of lemon liqueur 1109–1116.
(Limoncello) by conventional maceration and a two-syringe system to illustrate rapid Verma, A. R., Vijayakumar, M., Mathela, C. S., & Rao, C. V. (2009). In vitro and in vivo
solid-liquid dynamic extraction. Journal of Chemical Education, 92(5), 911–915. antioxidant properties of different fractions of Moringa oleifera leaves. Food and
Nouman, W., Anwar, F., Gull, T., Newton, A., Rosa, E., & Domínguez-Perles, R. (2016). Chemical Toxicology, 47, 2196–2201.
Profiling of polyphenolics, nutrients and antioxidant potential of germplasm's leaves Vongsak, B., Sithisarn, P., & Gritsanapan, W. (2013). Simultaneous determination of
from seven cultivars of Moringa oleifera Lam. Industrial Crops and Products, 83, crypto-chlorogenic acid, isoquercetin, and astragalin contents in Moringa oleifera leaf
166–176. extracts by TLC-densitometric method. Evidence-Based Complementary and Alternative
Olabode, Z., Akanbi, C. T., Olunlade, B., & Adeola, A. A. (2015). Effects of drying tem- Medicine. (Volume 2013, Article ID 917609, 7 pages).
perature on the nutrients of Moringa (Moringa oleifera) leaves and sensory attributes Vongsak, B., Sithisarn, P., Mangmool, S., Thongpraditchote, S., Wongkrajang, Y., &
of dried leaves infusion. Direct Research Journal of Agriculture and Food Science, 3, Gritsanapan, W. (2013). Maximizing total phenolics, total flavonoids contents and
117–122. antioxidant activity of Moringa oleifera leaf extract by the appropriate extraction
Oyeyinka, A. T., & Oyeyinka, S. A. (2018). Moringa oleifera as a food fortificant: Recent method. Industrial Crops and Products, 44, 566–571.
trends and prospects. Journal of the Saudi Society of Agricultural Sciences, 17, 127–137. Wang, T., Li, Q., & Bi, K. (2018). Bioactive flavonoids in medicinal plants: Structure,
Pereira, G. A., Molina, G., Arruda, H. S., & Pastore, G. M. (2017). Optimizing the activity and biological fate. Asian Journal of Pharmaceutical Sciences, 13, 12–23.

327