Subscriber access provided by SETON HALL UNIV
Article
Olive Fruit Phenols Transfer, Transformation and Partition Trail During Laboratory-Scale Olive Oil Processing Tina Jerman Klen, Alenka Golc-Wondra, Urska Vrhovsek, Paolo Sivilotti, and Branka Mozetic Vodopivec J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/jf506353z • Publication Date (Web): 20 Apr 2015 Downloaded from http://pubs.acs.org on April 27, 2015
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Agricultural and Food Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 30
Journal of Agricultural and Food Chemistry
Olive Fruit Phenols Transfer, Transformation and Partition Trail During LaboratoryScale Olive Oil Processing
Tina Jerman Klen1, Alenka Golc Wondra2, Urška Vrhovšek3, Paolo Sivilotti1, Branka Mozetič Vodopivec1*
1
Wine Research Centre, University of Nova Gorica, Glavni trg 8, 5271 Vipava, Slovenia 2
Centre for Validation Technologies and Analytics, National Institute of Chemistry, Hajdrichova 19, Ljubljana 1000, Slovenia
3
Department of Food Quality and Nutrition, Research and Innovation Centre, Fondazione Edmund Mach (FEM), Via E. Mach, 1 38010 S. Michele all'Adige (TN), Italy
* corresponding author: tel.: + 38659099701, e-mail:
[email protected] 1
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
1
ABSTRACT
2
This work is the most comprehensive study on quantitative behaviour of olive fruit phenols
3
during olive oil processing, providing insight into their transfer, transformation and partition
4
trail. In total, 69 phenols were quantified in 6 olive matrices from a three-phase extraction line
5
employing ultra high pressure liquid chromatography – diode array detection analysis.
6
Crushing had bigger effect than malaxation in terms of phenolic degradation and
7
transformation, resulting in several new evolutions of respective derivatives. The peel and
8
pulp together confined 95% of total fruit phenols, while stone only 5%. However, only 0.53%
9
of all ended-up in olive oil, nearly 6% in wastewater and 48% in pomace. Secoiridoids were
10
the predominant class in all matrices, though represented by different individuals. Their
11
partition behaviour was rather similar to other phenolic classes, where with few minor
12
exceptions only aglycones were partitioned to the oil, while other glycosides were lost with
13
the wastes.
14 15
Keywords: phenols; olive oil processing; transfer, partition; transformation; U(H)PLC; DAD,
16
crushing; malaxation
17
2
ACS Paragon Plus Environment
Page 2 of 30
Page 3 of 30
18
Journal of Agricultural and Food Chemistry
INTRODUCTION
19
Olive phenols have drawn increased attention over the past few decades owing to diverse
20
range of bioactivities such as antioxidant, antimicrobial, anticancer and others,1,2 assigning
21
them as one of the most valuable and promising dietary compounds. On the other hand, these
22
constituents have been recognized as serious environmental pollutants with several proven
23
toxicities against terrestrial and aquatic organisms, causing troubles in olive industry
24
sustainable development.3 In particular, olive oil processing has been associated with a huge
25
loss of valuable phenolic compounds,5,6 owing to biological (e.g. fruits enzymatic level),
26
technological (e.g. malaxation conditions) and other limitative factors (e.g. phenols
27
liposolubility problems etc.) facing industry with several challenges, still waiting to be
28
resolved.
29
Olive phenols entailed in olive oil processing encompass those originally present in olive
30
fruits and those newly formed via various chemical and/or enzymatic reactions. The initial
31
fruit phenolic content derived from either peel, pulp and/or stone present the available pool of
32
phenols that could end-up in one or all of the final products of olive oil processing, i.e. oil and
33
wastes (pomace and wastewater), however, the form and extent in which they reach them is
34
still poorly understood and scarcely investigated though significant from all – the health,
35
economic and ecological perspectives. In 2002, Rodis et al.,5 reported that only 1−2% of the
36
available fruit phenols are transferred to olive oil, while the rest (98%) are lost with wastes.
37
Likewise, our experimental data from previous industrial-scale research6 revealed their low
38
partition, since only 0.3−1.5% of the initial fruit phenolic content was found in oil, while the
39
rest was destructed or lost with the wastes formed. The transfer of phenols from fruits to
40
paste7–9 and from paste10–12 to the final products (oil and wastes) has also been quantified in
41
other reports, but the results presented only for a limited range of phenols, restricting an
42
adequate study of their degradation and/or transformation pathway.
3
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
43
As olive oil processing continues to grow and the world becomes more diet- and eco-
44
conscious, the investigations of olive fruit phenols entailed in olive oil production will likely
45
to expand. In response to such upcoming trends, the present study was undertaken as one of
46
the early attempts to quantify their transfer, transformation and partition trail as only such
47
knowledge can facilitate manipulation of their levels and occurrence in the food and waste
48
matrices.
49
The work reported here is, in fact, a part of a wider investigation exploring the fate of olive
50
fruit phenols during olive oil production using a laboratory-scale three-phase extraction line.
51
In our previous study,13 we qualitatively trailed their fate through the processing, while in this
52
we provide a further insight into their quantitative behaviour, whereas the potency of their
53
partition regulation has been investigated in an ongoing research.
54
Quantification phenol analyses of 6 olive oil production matrices from a controlled trial
55
allowed us to comprehensively study quantitative partitions/transfers or transformations of
56
phenols in the olive oil process without the impact of different plant material, different sample
57
preparation/extraction procedures and chromatographic conditions which can hamper such
58
phenol studies. To our best knowledge, this work is the most comprehensive study on a
59
quantitative behaviour of olive fruit phenols entailed in olive oil production, providing insight
60
into the origin, degradation and evolution of each throughout the processing, with novel
61
transfer/partition calculations and transformative relationships established.
62 63 64 65
MATERIALS AND METHODS Reagents and standards. All reagents and standards were obtained and prepared as described.13
66
Sampling and samples pre-treatment. Olive fruits of Istrska belica cv. were harvested in
67
Gradno (Goriška Brda, Slovenia), stored overnight and processed to olive oil the next day
4
ACS Paragon Plus Environment
Page 4 of 30
Page 5 of 30
Journal of Agricultural and Food Chemistry
68
using a laboratory-scale Abencor olive mill (Seville, Spain). Prior to that, a lot of fruits were
69
randomly chosen as a representative starting fruit material, weighed and frozen with liquid
70
nitrogen before subjected to freeze-drying. Then, fruits were manually de-stoned and re-
71
balanced in order to obtain the pulp/stone average mass ratio, ground separately into
72
homogeneous powder with a liquid nitrogen and stored at −25 °C until analysis. Other olive
73
oil process-derived matrices (pomace, wastewater and oil) were sampled at the end of
74
processing with an exception of paste, which was collected immediately after fruits crushing.
75
The samples were prepared as previously described,14−16 where de-stoned fruits, stones, paste
76
and pomace were frozen with liquid nitrogen, freeze-dried, homogenised and stored at −25 °C
77
until analysis, while the wastewater was primarily acidified (HCl, pH = 2.0) and defatted with
78
n-hexane prior to freeze-drying and storage at −25 °C. The olive oil sample was solely stored
79
in the dark at room temperature until extracted. The dry matter measurements of freeze-dried
80
samples were performed gravimetrically,17 providing the basis for phenols transfer/partition
81
calculations.
82
Olive oil processing. A laboratory-scale olive oil processing trial was carried out as
83
described elsewhere.13 First, olives were ground to a paste by the hammer mill and
84
homogenised in order to reduce the potential differences in the starting fruit material. Then,
85
the paste (∼700 g) was put into a metallic pitcher and 100 mL of water (25 °C) added to
86
improve its rheology. Afterword, the paste was malaxed in thermo-malaxer (30 min, 25 °C)
87
and centrifuged (1400 g, 1 min) in order to obtain the three final products, i.e. oil and two
88
wastes (wastewater and pomace). While the yield of pomace, which remained in a centrifuge
89
was calculated, the masses of two liquid phases (oil + wastewater) separated by decantation
90
process were balanced in order to assess the mass balance of the process trail. All the
91
procedure was performed in duplicate.
5
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
92
Phenols extraction. De-stoned fruits, stones, paste, pomace and wastewater. Phenols were
93
extracted at room temperature according to a previously published ultrasound-assisted solid
94
liquid extraction (USLE) method,14,15 where a freeze-dried sample (1.5 g) was sonicated (3 ×
95
20 min) with 25 mL of methanol. The homogenates of each extraction step were centrifuged
96
(2515 g, 5 min) and combined supernatants diluted with methanol to 100 mL.
97
Olive oil. Extraction (at room temperature) was carried using ultrasound-assisted liquid
98
liquid extraction (US-LLE) method previously described,16 where a 10 g of olive oil was
99
dissolved in n-hexane (10 mL) and sonicated (3 × 10 min) with pure methanol (5 mL). The
100
homogenates of each extraction step were centrifuged (2515 g, 2 min), combined and defatted
101
with a freeze-based fat precipitation (for 20 min at − 80 °C). Afterward, the remaining extract
102
was reconstituted to 25 mL with methanol and stored in the screw-capped dark glass
103
containers at −25 °C until analysis.
104
U(H)PLC-DAD analysis. Prior to analysis, the aliquots of extracts were rotary
105
concentrated 10-fold and the dry-residue re-dissolved in a mixture of U(H)PLC eluents A
106
(water: acetic acid = 95:5, v/v) and B (methanol) (A:B = 90:10; v/v). The extracts were
107
filtered through 0.22 µm/13 mm PVDF filters from Carl Roth GmbH+Co. (Düren, Germany)
108
and immediately analysed by U(H)PLC-DAD analysis.
109
A Dionex Ultimate 3000 U(H)PLC liquid chromatograph (Thermoscientific, CA, USA)
110
equipped with UV-Vis diode-array (DAD) detector was used for the quantification. The
111
extracts of 10 µL (except for oil; 20 µL) were injected and the DAD signals were recorded at
112
280, 320 and 365 nm, presenting a compromise for individual phenolic class detection.
113
Chromatographic separation was achieved by using gradient elution on Kinetex PFP column
114
(2.6 µm, 100 mm × 4.6 mm) attached to a PFP security guard (2.1 mm × 4.6 mm) both from
115
Phenomenex (Sydney, Australia) under the analysis conditions previously described.13
6
ACS Paragon Plus Environment
Page 6 of 30
Page 7 of 30
Journal of Agricultural and Food Chemistry
116
Flavonols were quantified at 365 nm, while simple phenols and secoiridoids at 280 nm with
117
an exception of unknown A, uknowns 408 MW, caffeoyl-6’-secologanoside and
118
comselogoside quantified at 320 nm along with the group of cinnamic acids and flavonoids.
119
When authentic standards were available, phenols were quantified based on their calibration
120
curves, whereas others were expressed as equivalents; hydroxytyrosol glucosides and
121
hydroxytyrosol acetate as hydroxytyrosol, tyrosol glucoside as tyrosol, all verbascoside
122
derivatives as verbascoside, luteolin rutinosides as luteolin-7-O-glucoside, luteolin-3’-O-
123
glucoside and luteolin-4’,7-O-diglucoside as luteolin-4’-O-glucoside, and all secoiridoids as
124
oleuropein with an exception of caffeoyl-6’-secologanoside expressed as caffeic acid, and
125
unknown A, unknowns 408 MW and comselogoside, which were expressed as p-coumaric
126
acid equivalents. The identification strategy of all is in depth described elsewhere.13
127
U(H)PLC method validation. Prior to quantification, the suitability of U(H)PLC-DAD
128
system was checked in terms of injector’s reproducibility and linearity using commercially
129
available standards.
130 131
The linearity of method was evaluated by serial dilution of standard stock solutions over broad concentration ranges using ten-point calibration curves.
132
The LODs and LOQs were calculated from y-intercept standard deviations (Sb) and slopes
133
(a) of calibration curves using signal-to-noise ratio criteria of 3.3 (LOD = 3.3 x Sb/a) and 10
134
(LOQ = 10 x Sb/a) in the concentration ranges close to LOQs expected for each phenolic
135
compound in prepared extracts (µg/mL): 0.33−1.67 (vanillin), 0.35−1.73 (vanillic acid),
136
0.33−1.65 (p-coumaric acid), 1.24−7.44 (caffeic acid), 1.44−5.76 (hydroxytyrosol), 1.40−3.93
137
(tyrosol), 1.00−2.99 (oleuropein), 0.39−0.97 (verbascoside), 0.24−1.20 (rutin), 0.23−1.17
138
(quercitrin), 0.25−1.23 (luteolin-4’-O-glucoside), 1.62−5.85 (apigenin), 0.48−2.87 (apigenin-
139
7-O-glucoside), 0.56−4.49 (luteolin), 1.26−3.78 (luteolin-7-O-glucoside) and 0.79−4.76
140
(pinoresinol). 7
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
141
Extraction efficiencies of USLE and US-LLE methods were checked prior to application to
142
olive matrices in terms of phenol yield recoveries obtained via consecutive five-step
143
extraction. The extracts of de-stoned fruits, stone, paste, pomace and wastewater were
144
submitted to a complete USLE method (3 × 20 min) using 25 mL of pure methanol. Then, the
145
remaining solid residue was re-extracted twice and analysed by U(H)PLC for a potential
146
phenols presence in the 4th and 5th extraction step, where the summation of phenol yields
147
from a five-step extraction was considered as 100%. Analogously, the efficiency of US-LLE
148
method (3 × 10 min, 5 mL 100% meOH) has been checked for the oil sample.
149
Phenols partition and transfer percentages. The phenols partition and transfer percentages
150
from fruits to paste and to final products (pomace, wastewater, oil) were calculated
151
considering the mass balance according to the equation (1):
152 153
%Pmatrix = (Pmatrix/Pfruit) × (mmatrix/mfruit) × 100
(Eqs 1)
154 155
where P is the concentration of individual phenolic compound in particular matrix expressed
156
in mg/kg per fruit FW and m the mass of selected matrix and fruit FW (g) in the press
157
experiment.
158 159
Additionally, the percentages of their increases/decreases during crushing and malaxation
160
were calculated using equations (2) and (3):
161 162
Increase/decrease during crushing is presented with parameter cc (in %):
163 164
cc = (Ppaste × 100/Pfruit ) – 100
(Eqs 2)
165 8
ACS Paragon Plus Environment
Page 8 of 30
Page 9 of 30
166
Journal of Agricultural and Food Chemistry
Increase/decrease during malaxation is presented with parameter cm (in %):
167 168
cm = ((Ppomace + Pwastewater + Poil) × 100/Ppaste ) – 100
(Eqs 3)
169
The initial fruit phenolic content was considered as the available (100%) pool derived from
170
de-stoned fruit and stone marked as input (Table 1), while the term TP (total phenols) refers to
171
a sum of U(H)PLC-DAD quantified phenols.
172
Statistical analysis. All results were expressed as means ± SD obtained from at least
173
triplicate analyses and tested by ANOVA or t-test (*, P < 0.05; **, P < 0.01; ***, P < 0.001;
174
n.s., not significant) using Costat Statistics Software 6.4 (CoHort Software, CA, USA). When
175
ANOVA was significant, means were separated using Tukey’s HSD test (P < 0.05).
176 177
RESULTS AND DISSCUSION
178
179
Method validation. The suitability of U(H)PLC-DAD system was satisfactory, where the
180
linearity of injector was always above 99% and the reproducibility of ten replicates has never
181
exceeded 0.5% of RSD, respectively.
182
The linearity of method was excellent with high correlation coefficients (R2) obtained for all
183
standards over their broad concentration ranges (µg/mL) tested as follows; vanillin (y =
184
0.6348x; R2 = 0.9933; 0.33−64.83), vanillic acid (y = 0.2899x; R2 = 0.9957; 0.35−67.26), p-
185
coumaric acid (y = 0.7863x; R2 = 0.9946; 0.33−64.21), caffeic acid (y = 0.6081x; R2 =
186
0.9982; 1.24−99.19), hydroxytyrosol (y = 0.1488x; R2 = 0.9970; 1.44−524.19), tyrosol (y =
187
0.0993x; R2 = 0.9937; 1.40−219.00), oleuropein (y = 0.0359x; R2 = 0.9963; 1.00−7953.12),
188
verbascoside (y = 0.2403x; R2 = 0.9991; 0.39−1013.12), rutin (y = 0.2368x; R2 = 0.9952;
189
0.24−312.40), quercitrin (y = 0.2329x; R2 = 0.9925; 0.23−121.44), luteolin-4’-O-glucoside (y
9
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
190
= 0.1775x; R2 = 0.9976; 0.25−896.56), apigenin (y = 0.2293x; R2 = 0.9988; 1.62−232.80),
191
apigenin-7-O-glucoside (y = 0.3362x; R2 = 0.9999; 0.48−274.56), luteolin (y = 0.2684x; R2 =
192
0.9928; 0.56−483.12), luteolin-7-O-glucoside (y = 0.2487x; R2 = 0.9962; 1.26−765.24),
193
pinoresinol (y = 0.1429x; R2 = 0.9975; 0.79−285.76).
194
The sensitivity of DAD detector was determined by calculating the LODs/LOQs for each
195
standard expressed in µg/mL and were as follows; 0.36/1.09 (vanillin), 0.23/0.69 (vanillic
196
acid), 0.22/0.67 (p-coumaric acid), 2.46/7.44 (caffeic acid), 2.55/7.74 (hydroxytyrosol),
197
1.75/5.29 (tyrosol), 1.57/4.76 (oleuropein), 0.19/0.58 (verbascoside), 0.03/0.01 × e-1 (rutin),
198
0.01/0.04 (quercitrin), 0.12/0.36 (luteolin-4’-O-glucoside), 1.20/3.87 (apigenin), 0.17/0.52
199
(apigenin-7-O-glucoside), 0.53/1.60 (luteolin), 0.16/0.49 (luteolin-7-O-glucoside) and
200
0.04/0.11 (pinoresinol).
201
The extraction methods for the phenol isolation from olive matrices entailed in olive oil
202
processing trial were adopted from our earlier reports performed as a preliminary step toward
203
a high-yielding TP analysis in fruits, wastewater and olive oil.14-16 However, prior to
204
application in this experiment and evaluation with U(HPLC)-DAD, their efficiencies had to
205
be re-checked owing to introduction of new olive matrices of richer quali- and quantitative
206
profiles and novel ones for which the recovery optimisations have not been yet performed
207
(paste, pomace and stone). Even so, our results confirmed that both USLE and US-LLE
208
extractions are efficient for the quantitative phenol analysis, where a three-step extraction
209
provided recoveries superior to 98% on average for all six matrices (US-LLE for the oil and
210
USLE for others), again testifying to high powerful US extracting abilities
211 212
Ouantification resuls. Table 1 presents a detailed insight into olive fruit phenols transfer,
213
transformation and partition trail behaviour during olive oil processing as affected by two 10
ACS Paragon Plus Environment
Page 10 of 30
Page 11 of 30
Journal of Agricultural and Food Chemistry
214
operational steps, i.e. crushing and malaxation (30 min/25 °C). The process itself has two
215
inputs, i.e. de-stoned fruit and stone, one middle/intermediae product (paste) and three
216
outputs, i.e. pomace, wastewater and oil, allowing to follow the origin, evolution and
217
disappearance of each phenolic compound throughout the processing. The identification of all
218
compounds presented in this study is described elsewhere. 13
219
Impact of crushing. The relative contribution of crushing step to a phenols reduction during
220
olive oil processing has been rarely quantified in the existing literature. Similarly, a little
221
scientific information is available on its role in their transformative process as the range of
222
phenols analysed so far has been limited.6-9,18 A direct comparison of olive fruits initial
223
phenolic composition (de-stoned fruit + stone) with the corresponding paste obtained directly
224
after crushing (t = 0) is the first approach to a such estimation. Looking quantitatively, a
225
significant TP drop was observed immediately after crushing (46%) confirming the previous
226
postulation of being the most critical step in the overall process.4 A similar TP lost (50−60%)
227
has already been reported before for industrial-scale trials.6,9 Although brief, this
228
technological operation has also induced several transformative changes arriving from the
229
mechanical mixing action, chemical and/or biochemical reactions, i.e. enzymatic and non-
230
enzymatic hydrolyses and oxidations.19 As seen from results, crushing implied a rise of simple
231
phenols and benzoic acids, but decreased the yields of other classes with an exception of
232
lignans, which were not involved/affected by milling process. At the level of individual
233
phenols, thirteen out of fifty-one initially quantified fruit phenols have irrevocably
234
disappeared after crushing (mainly those from stone) and eight newly appeared in the paste,
235
indicating their technologically induced formation and/or release. The phenol profile of olive
236
paste thus seems to be a result of all, the transfer, liberation and transformation phenomena
237
with none, partial or total hydrolysis/degradation of olive fruits native phenols leading to a
11
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
238
formation of new respective derivatives. A visual comparison of fruits’ and paste’s phenolic
239
profiles (Figure 1) reflects some of the potential interconversion changes discussed below.
240
The highest decrease was observed for the main olive fruit secoiridoid glucosides, i.e.
241
oleuropein, demethyloleuropein and ligstroside, which is in line with previous reports.6,7−9
242
While the content of oleuropein decreased for 94%, its demethylated form was almost
243
completely degraded (traces of demethyloleuropein), whereas ligstroside could not be
244
detected by DAD anymore. According to the mechanism of Servili et al.,4 these glucosides
245
can be transformed into their respective aglycones; primarily to oleuropein or ligstroside
246
aglycones and further to their decarboxymethylated forms, i.e. 3,4-DHPEA-EDA or p-HPEA-
247
EDA. Indeed, both of the potential (demethyl)oleuropein transformants were found in olive
248
paste after crushing; their first interconversion product eluted as three (oleuropein aglycone
249
isomers), while the second as two peaks (3,4-DHPEA-EDA isomers), together accounting 35
250
and 12% of fruits initial demethyloleuropein + oleuropein content together. This could
251
indicate that crushing accelerated an enzymatic degradation of both glucosides, which already
252
began in the fruit and would likely to progress during maturation. Such an assumption is
253
supported by the previous MS trace detection of oleuropein aglycones in de-stoned fruits13
254
and the fact that 3,4-DHPEA-EDA has already been found in the fruits before.20 Analogously,
255
these transformants could also arrive from oleuropein diglucoside isomer 2, whose complete
256
degradation was observed after crushing, while the rise of its isomer 1 (+ 64%) so far
257
constitutes unexplained and warrants a further investigation. Nevertheless, the transformation
258
of these native fruit secoiridoid glucosides could also yield some other oleuropein structurally
259
correlated phenols whose yields have increased after crushing such as dihydro-oleuropein
260
isomer 1 or others of lower molecular weight (MW) phenols. Indeed, there are many cleavage
261
products possible upon the degradation of oleuropein, demethyloleuropein and ligstroside, of
262
which some are illustrated in Figure 2.
12
ACS Paragon Plus Environment
Page 12 of 30
Page 13 of 30
Journal of Agricultural and Food Chemistry
263
By contrast, the degradative mechanism of ligstroside was more ambiguous as none of its
264
two potential transformants (ligstroside aglycones and p-HPEA-EDA) could be detected in
265
paste by DAD, but instead could easily be quantified in the oil at the end of the processing.
266
The relationship that quantitatively explains such vague interconversion remains unclear and
267
may indicate that; i) ligstroside was completely transformed into both, but the complexity of
268
paste’s structure restricted their detection in paste as its composition immediately after
269
crushing is known to strongly differ from that after malaxation,21 or ii) these two
270
transformants could have formed the complexes with polysaccharides released only after
271
malaxation,22 or (less likely) iii) ligstroside was completely oxidised and/or transformed into
272
other products, while its respected aglycones found in olive oil are the hydrolysis products of
273
others, but structurally related compounds.
274
However, beyond the knowledge of fruits’ prime secoiridoid glucosides behaviour, the
275
degradative mechanism of stone’s main secoiridoid glucosides i.e. nüzheonide and its
276
esterified forms with methoxy group and 11-methyl oleoside has not been yet established.
277
Interestingly, only two out of ten quantified representatives were increased upon olives
278
crushing (nüzhenide isomer 2 for 21% and nüzhenide 11-methyl oleoside isomer 4 for 258%),
279
most likely due to improved release and/or the mill prompted hydrolysis of higher MW
280
phenols such as nüzhenide (di)esters giving rise to their formation. All the other nüzhenide
281
representatives have totally diminished throughout the milling operation due to either low
282
oxidation resistance and/or crushing-induced degradation. Nevertheless, based on their
283
chemical structures they could all yield a tyrosol glucoside, whose significant rise in paste
284
(134%) should be marked after crushing.
285
Crushing also implied a rise of other simple phenols (Figure 1a, Figure 2a); hydroxytyrosol
286
has increased for 170% owing to degradation of hydroxytyrosol-containing compounds, while
287
tyrosol for 107% due to the hydrolysis of tyrosol-containing phenols. This could have
13
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
288
occurred via cleavage of ester bond by the action of endogenous esterases splitting secoiridoid
289
glucosides or their aglycones (Figure 2c), giving rise to elenolic acid glucoside, oleoside or
290
elenolic acid (Figure 2b). Until now, the release of simple phenols from aglycones was
291
considered to ensue only during oil storage23 and never during course of its production.
292
Alternatively, hydroxytyrosol could also be released from the fruits native hydroxytyrosol
293
glucoside (-29%) and/or verbascoside (-31%) by the activity of glucosidases, analogously as
294
tyrosol from nüzhenide and its esters. Owing to such complexity, it was not possible to assign
295
their exact origin as no obvious increase of the corresponding cleavage compounds could be
296
observed in olive paste. Another remarkable fact is also a distinct behaviour of the two
297
hydroxytyrosol glucosides displaying rise (70%, hydroxytyrosol glucoside-1-β-glucoside) and
298
a drop (29%, hydroxytyrosol glucoside) after olives crushing. Such a discrepancy clearly
299
indicates to their different chemical structure and for the first time experimentally supports
300
(during course of olive oil production) what has only been proposed before;1 hydroxytyrosol
301
glucoside-1-β-glucoside is a hydrolysis product of verbascoside enhanced by both, the
302
crushing and malaxation, while the formation of hydroxytyrosol glucoside appears not be
303
technologically prompted; instead it rather contributes to its degradation.
304
The appearance of vanillic acid in olive paste after crushing raises an issue of its origin. The
305
fact that it was absent in both input matrices, but present in all olive oil process-derived
306
products suggests that is a formation product of the technological process. Possible
307
explanations include a crushing-induced hydrolysis of lignin from stone once already proved
308
to yield vanillic acid under acid stem explosion,24 although there is always a possibility that a
309
lab-milling of stone in a sample preparation was not efficient as the mill crushers of Abencor
310
system hampering its expected detection in this fruit compartment. Interestingly, its
311
quantitative yield was further unaffected by the malaxing conditions, which seems to only
312
facilitated its full distribution among the final products. By contrast, vanillin was detected in
14
ACS Paragon Plus Environment
Page 14 of 30
Page 15 of 30
Journal of Agricultural and Food Chemistry
313
olive stone, but not in the corresponding paste, which could be attributed to a brief duration of
314
crushing step, suffering less tissue damage and hence a limited release immediately after it.
315
The subsequent malaxation apparently induced its extractability, which allowed its
316
quantification in olive oil, though in a much lower amount in respect to its original content in
317
stone (3%).
318
Among cinnamic acids, only verbascoside has diminished after fruits crushing (-31%),
319
whereas all of its derivatives have considerably increased (>150%, Figure 1b). It is of interest
320
to add that their accumulation has further proceeded also during malaxation, thus it is very
321
likely that verbascoside derivatives had originated from any of the fruits’ native unknown
322
compound(s) hydrolysed upon crushing and malaxation. Moreover, their initial presence in
323
olive fruits suggests that the technology has only hasten the biosynthetic inter-conversion that
324
would likely occur also at the fruit level, however, such a hypothesis undoubtedly warrants a
325
further investigation, and is beyond the scope of the present study.
326
Crushing also contributed to a loss of glycosylated flavonoids with an exception of luteolin
327
3’-O-glucoside, which seemed to be resistant to any or all of enzymatic and/or non-enzymatic
328
degradations, transformations or oxidations. The appearance of two new flavonoidal
329
aglycones in olive paste, i.e. luteolin and apigenin (Figure 1c), indicates that some of them
330
were hydrolysed by the endogenous glucosidases, released and/or activated by crushing, but a
331
firm relationship could only be established for apigenin whose formation in olive paste was
332
collaborated with a drop of apigenin-7-O-glucoside. Luteolin, on the other hand, may have
333
arisen from any or all of the six luteolin glycosides, rutin and/or quercitrin.
334
In spite of several quali- and quantitative induced changes by crushing, a small portion of
335
phenols remained unaffected, including luteolin-3’-O-glucoside, comselogoside and caffeoyl-
336
6’-secologanoside. A remarkable fact is that all of their yields have further remained constant
15
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
337
also during malaxation, indicating to their high technological resistance, which could have a
338
wide potential in a forward design of TP enriched food products.
339
Impact of malaxation. Although phenols yield behaviour during course of malaxation has
340
not been separately monitored, their levels in the output matrices yet permitted a reliable
341
insight into their further transformation/partition trail from paste to the final products,
342
especially if considered that the time of paste’s span in the centrifuge is too short to allow
343
important modifications happen.
344
In general, the behaviour of phenols during malaxation continued the trend initiated by the
345
crushing, though quantitatively in a lower extend; a positive yield rises kept an increasing
346
trend due to improved releases and/or transformative reactions, while the yield losses were
347
followed by a further drops owing to enzymatic and/or non-enzymatic hydrolyses and
348
oxidations. However, there were also few exceptions; for example luteolin rutinoside isomer 1
349
showed a rise after initial drop, while some of the potential transformants have decreased, i.e.
350
elenolic acid glucoside isomer 1, 3,4-DHPEA-EDA isomer 1, nüzhenide isomer 2, oleuropein
351
aglycone isomer 1 and nüzhenide 11-methyl oleoside isomer 4. This could indicate that the
352
hydrolysis of their parent compounds has finished and their oxidative degradation prevailed
353
due to operative conditions and/or enzymatic activities. By contrast, some other potential
354
transformants (dihydro-oleuropein isomer 1 and oleuropein aglycone isomer 3) remained
355
rather unaffected, similarly as most of the flavonoidal representatives. Moreover, the
356
evolution of some new phenols could be evidenced in the final products; both representatives
357
of lignans and some of secoiridoids (all unknown 484 MW isomers, p-HPEA-EDA,
358
ligstroside aglycones and acetal of 3,4-DHPEA-EDA) have appeared in the oil or in the
359
wastewater, while not in pomace, though some traces could still be previously detected by MS
360
as well.13 It is thus possible to deduce that only malaxation conditions were efficient enough
361
to induce their formation and/or release due to prolonged mixing and/or enzymatic actions.
16
ACS Paragon Plus Environment
Page 16 of 30
Page 17 of 30
Journal of Agricultural and Food Chemistry
362
However, there is always a possibility that the extraction method for these compounds was
363
not equally efficient for all matrices owing to structural specifies, requiring other approaches
364
as previously demonstrated for lignans in fruits.25
365
Looking from a TP yield perspective, no losses could be observed during malaxation at 30
366
min/25 °C conditions, indicating to a balanced degradation/formation equilibrium of the
367
phenolic compounds quantified. In other words, the available pool of phenols detected in
368
olive paste after crushing could also be quantified at the end of process via TP sum of output
369
products (∼54%). This is somewhat in line with previous report,9 demonstrating no significant
370
losses of phenol constituents during malaxation process, though this operation has often been
371
associated with major quantitative lost.8 However, as further evident from results, only 0.53%
372
of the available fruit phenols have ended-up in olive oil and nearly 6% in the wastewater,
373
while others have remained entrapped in the solid (48.12%). Such partition distribution is
374
rather different from our earlier industrial-scale results6 (0.3−1.2% oil; 38.2−46.2%
375
wastewater; 4.5−47.4% pomace), but not surprising as the latter is govern by the quantity of
376
final products formed. In a present lab-scale study, pomace was clearly the predominant by-
377
product formed (82.3%), while the yielding indexes of liquid matrices were much lower, i.e.
378
8.9% for oil and 8.8% for the wastewater, most likely due to a less efficient phases separation
379
system in the Abencor mill. A similar components’ yield distribution has been observed
380
elsewhere.11
381
A high retention of phenols in pomace further suggests that the process conditions applied
382
(30 min/25 °C) has not fully induced their transfer to liquid phases, yet constituting a
383
challenge for some potential improvements. Our results also confirmed that a major
384
proportion of fruit phenols is indeed lost with wastes, but some also owing to their
385
technological destruction. This, however, is not in full agreement with the available results in
386
the literature,5 showing no TP losses during olive oil processing; but rather their unsuited 17
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
387
yield distribution among the final products, i.e. 1−2% oil, 53% wastewater and 45% pomace.
388
However, it is rather uncommon that processing would quantitatively yield all the phenols
389
confined in the initial fruit material (100% TP pool), especially not due to a well-known
390
phenomenon of their oxidative catabolism.4
391
Partitioning model. The majority of phenols have originated from the pulp and/or peel,
392
together accounting 95% of the fruits initial TP content, while stone confined only a minor
393
fraction (∼5%). Secoiridoids were the predominant class in both input matrices and
394
maintained its prominent role also in the process-derived matrices, though represented by
395
different individuals. With an exception of olive stone, the class distribution in solid matrices
396
has decreased in the order; secoiridoids > flavonoids > cinnamic acids > simple phenols. By
397
contrast, in the wastewater, simple phenols constituted the second largest class evidencing
398
their hydrophilic nature, analogously as lignans in olive oil showing high tendency for the oily
399
matrix. Interestingly, none of the fruit flavonoid glycosides were transferred to the oil above
400
their trace amounts, but instead were largely confined in both of the waste matrices. Similarly,
401
all cinnamic acids from fruits were lost with the by-products, though one has scarcely been
402
transferred to the oil as well, i.e. p-coumaric acid. The partition behaviour of secoiridoids was
403
rather similar to flavonoids, where apart from the two known glucosides (secologanoside and
404
elenolic acid glucoside isomer 2), only aglycones were partitioned to the oil, while the other
405
glucosides have ended-up in the wastes (e.g. comselogoside and caffeoyl-6’-secologanoside).
406
Such partitioning model is quite similar to a previously reported at industrial-scale level using
407
different starting fruit material,6 which indicates that their qualitative distribution pattern
408
among the final products is rather common than unique.
409
The research interests to redesign the current technological approaches to provide higher
410
extractability of olive oils with richer phenol yields and of lower ecological footprints have
411
been great in recent years and are still significant due to increasing demands for functionally
18
ACS Paragon Plus Environment
Page 18 of 30
Page 19 of 30
Journal of Agricultural and Food Chemistry
412
and sustainability in olive oil industry. The results of present study have provided us a deeper
413
insight into phenols quantitative behaviour during laboratory-scale olive oil processing, their
414
tendency for individual matrices and resistance to the technologically induced modifications,
415
being important in the recognition of their quality and value to mankind. Moreover, they
416
provided us a better understanding of the relationship between the initial and final products,
417
which could benefit in any further industrial-scale trials.
418 419
SUPPORTING INFORMATION
420 421
Supporting Information Available: U(H)PLC-DAD phenolic profile of: olive de-stoned fruit,
422
olive stone, olive paste, olive pomace, olive mill waste water and olive oil extracts monitored
423
at 280, 320 and 365 nm and results of additional statistical evaluation of the olive oil process;
424
ANOVA analyses of the Input (Fruit + Stone), Intermediate (Paste) and Output (OMWW +
425
Pomace + Olive Oil) Phenol Contents (mg/kg of Fruit FW) and Tukey's HSD separation of
426
means (p