Accepted Manuscript Accelerated solvent extraction of carotenoids from: Tunisian Kaki (Diospyros kaki L.), peach (Prunus persica L.) and apricot (Prunus armeniaca L.) Khalil Zaghdoudi, Steve Pontvianne, Xavier Framboisier, Mathilde Achard, Rabiga Kudaibergenova, Malika Ayadi-Trabelsi, Jamila Kalthoum-cherif, Régis Vanderesse, Céline Frochot, Yann Guiavarc’h PII: DOI: Reference:
S0308-8146(15)00444-6 http://dx.doi.org/10.1016/j.foodchem.2015.03.072 FOCH 17319
To appear in:
Food Chemistry
Received Date: Revised Date: Accepted Date:
16 October 2014 18 March 2015 20 March 2015
Please cite this article as: Zaghdoudi, K., Pontvianne, S., Framboisier, X., Achard, M., Kudaibergenova, R., AyadiTrabelsi, M., Kalthoum-cherif, J., Vanderesse, R., Frochot, C., Guiavarc’h, Y., Accelerated solvent extraction of carotenoids from: Tunisian Kaki (Diospyros kaki L.), peach (Prunus persica L.) and apricot (Prunus armeniaca L.), Food Chemistry (2015), doi: http://dx.doi.org/10.1016/j.foodchem.2015.03.072
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Accelerated solvent extraction of carotenoids from: Tunisian Kaki (Diospyros kaki L.),
2
peach (Prunus persica L.) and apricot (Prunus armeniaca L.)
3 4
Abbreviated running title suggested:
5
Accelerated solvent extraction (ASE) of carotenoids from kaki, peach and apricot.
6 7
Khalil Zaghdoudi 1-3, Steve Pontvianne 1, Xavier Framboisier 1, Mathilde Achard 2,
8
Rabiga Kudaibergenova 1, Malika Ayadi-Trabelsi 3, Jamila Kalthoum-cherif 3, Régis
9
Vanderesse 2, Céline Frochot 1, Yann Guiavarc’h 1*
10 Authors are affiliated to:
11 12
1
Laboratoire Réactions et Génie des Procédés (LRGP), UMR 7274, Université de Lorraine, ENSIC, 1 rue Grandville, 54001 Nancy, France
13 14
2
Lorraine, ENSIC, 1 rue Grandville, 54001 Nancy, France
15 16 17
Laboratoire de Chimie Physique Macromoléculaire (LCPM), FRE 7568, Université de
3
Laboratoire d'Application de la Chimie aux Ressources et Substances Naturelles et à l'Environnement, Faculté des Sciences de Bizerte, Université de Carthage, Tunisie
18 19
*Author to whom correspondence should be addressed
20
: +33 (0)3 83 17 51 90
21
Fax: +33 (3)83 32 29 75
22
:
[email protected] 23 24
Short title: “Accelerated solvent extraction of fruits carotenoids”
25 26 27
1
28
Abstract: 150 words.
29
Extraction of carotenoids from biological matrices and quantifications remains a difficult task.
30
Accelarated Solvent Extraction was used as an efficient extraction process for carotenoids extraction
31
from three fruits cultivated in Tunisia: Kaki ( Diospyros kaki L.), peach (Prunus persica L.) and apricot
32
(Prunus armeniaca L.). Based on a Design of Experiment (DoE) approach, and using a binary solvent
33
consisting of methanol and tetrahydrofurane, we could identify the best extraction conditions as
34
being 40°C, 20:80 (v:v) methanol/tetrahydrofuran and 5 min of extraction time. Surprisingly and
35
likely due to the high extraction pressure used (103 bars), these conditions appeared to be the best
36
ones both for extracting xantophylls such as lutein, zeaxanthin or β-cryptoxanthin and carotenes
37
such as β-carotene, which present quite different polarities. Twelve surface responses were
38
generated for lutein, zeaxanthin, β-cryptoxanthin and β-carotene in kaki, peach and apricot. Further
39
LC-MS analysis allowed comparisons in carotenoids profiles between the fruits.
40 41
Keywords: carotenoids, accelerated solvent extraction (ASE), Design of Experiment (DoE), kaki,
42
peach, apricot.
43
2
44
1. Introduction
45
Over the last decades, the interest for natural & bioactive compounds such as carotenoids grew up,
46
driven by the increase of consumers’ demand for healthy diets. Next to their coloring properties
47
applied in food and cosmetic industry, carotenoids are involved in the prevention of several diseases
48
such as cancers, age-related macular degeneration, cataracts, cardiovascular diseases and other
49
diseases related to low immune function (Perera & Yen, 2007). They provide photoprotection to the
50
eye skin and photosynthetic organisms through the quenching of singlet oxygen and damaging of
51
free radicals (Böhm, Edge & Truscott, 2012; Jomova & Valko, 2013). It is therefore not surprising that
52
the current market value of commercially used carotenoids was estimated at nearly $1.2 billion in
53
2010, with a chance to grow up to $1.4 billion in 2018 and a compound annual growth rate of 2.3%
54
(BCC Research LLC, 2011). Except for canthaxanthin, whose use for colorization of salmonid fish and
55
shrimps is likely to be strongly regulated or even forbidden in Europe, every other top 10 carotenoids
56
markets (β-carotene > lutein > astaxanthin > capsanthin> annatto > canthaxanthin > lycopene > β-
57
apo-8-carotenal > zeaxanthin > β-apo-8-carotenal ester) should increase by 2018. β-carotene, lutein
58
and astaxanthin represent more than 80 % of the market value (BCC Research LLC, 2011).
59
Carotenoids are lipid soluble pigments produced as secondary metabolites in fruits, vegetables,
60
algae, fungi and some bacteria. Animal are incapable of carotenoids biosynthesis and hence depend
61
on dietary carotenoids, which are more or less absorbed after their ingestion and, for some of them,
62
converted to provitamin A. In food, carotenoids are generally C40 tetraterpenoids formed from eight
63
C5 isoprenoid units joined head-to-tail, except at the center where a tail-to-tail linkage reverses the
64
order, resulting in a symmetrical molecule (Rodriguez-Amaya & Kimura, 2004; Ibanéz, Herrero,
65
Mendiola & Castro-Puyana, 2012). Carotenoids’ attractive color (red, yellow or orange) is due to their
66
extended conjugated double-bond system, which constitutes the light absorbing chromophore that
67
provides the visible absorption spectrum that serves as a basis for their identification &
68
quantification. They can be divided in two groups: oxygenated carotenoids, called xanthophylls, and
3
69
non-oxygenated carotenoids, called carotenes. To date, about 600 carotenoids have been identified
70
in nature and humans have access to about 40 carotenoids through their diet, mainly through fruits
71
and vegetables (Fernandez-Garcia, Carvajal-Lérida, Jarén-Galan, Garrido-Fernandez, Pérez-Galvez &
72
Hornero-Méndez, 2012). Among these carotenoids, β-carotene, α-carotene, lutein, zeaxanthin, β-
73
cryptoxanthin and lycopene are the most found in human diet with a good dietary intake and it was
74
evaluated that 80-90 % of their intake was coming from fruits and vegetables in developed countries
75
(Maiani et al., 2009).
76
Many investigations on carotenoids were undertaken in order to better understand their chemistry
77
and the health benefits due to the biological activity of these secondary metabolites. These studies
78
were based on and contributed to the development of various extraction and identification methods
79
of carotenoids and their related isomers in food matrices such as fruits & vegetables but also algae
80
and some bacteria (Amorim-Carrilho, Cepeda, Fente & Regal, 2014). As lipophilic compounds,
81
carotenoids are insoluble in water and show a high solubility in organic solvents. Carotenes are for
82
instance highly soluble in petroleum ether, hexane and toluene while xanthophylls dissolve better in
83
methanol and ethanol (Rodrigez-Amaya, 2001).
84
Over the last 50 years many efforts were put on the development of improved carotenoids
85
extraction processes as well as on their better identification through various analytical methods.
86
However, reproducible extraction/quantification of such pigments from various & complex biological
87
food matrices still remains a complex, time and cost demanding task (Amorim-Carrilho, Cepeda,
88
Fente, Regal, 2014; Taylor, Brackenridge, Vivier & Oberholster, 2006; Kimura & Rodriguez-Amaya,
89
2002). Carotenoids are relatively stable in the matrix, but in solution they may be very sensitive to
90
light, heat, acid or oxygen exposure and they can undergo isomerization during extraction/analysis
91
(Dias, Oliveira, Camoes, Nunes, Versloot & Hulshof, 2010; Zerlotti Mercadante, 1999). In a general
92
manner, short extraction time and protection from light and oxygen are therefore necessary for
93
reliable quantifications. Carotenoids can be found in their free form (carotenes) or in a more stable
4
94
fatty-acid esterified form (xanthopylls). These compounds are frequently analyzed after
95
saponification, which is an extraction step aimed at removing chlorophylls and lipids to release
96
carotenoids in a clean preparation for analysis, free from conjugated forms, fatty acids and lipids that
97
make the chromatography separation difficult (Murillo et al., 2013).
98
As far as possible, and due to their moderate to high hydrophobicity, conventional extractions of
99
carotenoids requires the use of organic solvents, which is costly, not environmentally friendly and
100
somewhat contradictory with the “green touch” associated with these natural compounds. Usually,
101
non-polar solvents, such as hexane or THF, are a good choice for extraction of non-polar (carotenes)
102
or esterified carotenoids, while polar solvents, such as methanol, ethanol and acetone, are more
103
appropriate for extraction of polar carotenoids (xanthophylls). Different extractions processes and
104
chromatographic methods to assay carotenoids in food matrices were the topic of several recent
105
reviews (Amorim-Carrilho, Cepeda, Fente, Regal, 2014; Arvayo-Enriquez, Mondaca-Fernandez,
106
Gortarez-Moroyoqui, Lopez-Cervantes & Rodriguez Ramirez, 2013; Mustafa & Turner, 2011). In a
107
brief and non-exhaustive way, three main categories of extraction technics are reported in the
108
literature, which are (i) the atmospheric liquid extraction with Soxhlet, maceration or ultrasound
109
methods (Mezzomo, Maestri, dos Santos, Maraschin, Ferreira, 2011), (ii) the PLE for pressurized
110
liquid extraction, also known as accelerated solvent extraction (ASE) and so called in this study
111
(Mustafa, Mijangos Trevino, Turner, 2012; Breithaupt, 2003), and (iii) the supercritical fluid
112
extraction, which is often based on the use of supercritical CO2 as solvent with slight use of organic
113
cosolvent such as ethanol (Guedes et al. 2013).
114
Liquid extraction technics at atmospheric pressure are often time-demanding but the Soxhlet
115
method, which is based on the recirculation of the organic solvent through the sample, using phase
116
transition (evaporation-condensation), can reduce the amount of organic solvent to be used. ASE, is
117
an interesting alternative to extractions at atmospheric pressure since it makes use of pressure
118
vessels containing the sample to be extracted and high constant pressure (often 103 bars), which
5
119
greatly facilitates the penetration of the extracting solvent, in a liquid form even when working
120
above its boiling temperature, through the food matrix. This drastically speeds up the extraction
121
process while decreasing the amount of solvent to be used (Mendiola, Herrero, Cifuentes & Ibanez,
122
2007). Automated laboratory scale equipments have been designed, which allow a perfect control of
123
pressure, temperature, time of extraction, and solvent composition together with the possibility to
124
program an extraction run for up to 24 samples placed in high pressure stainless-steel vessels where
125
they are protected from light and oxygen. Finally, the increasing use of supercritical CO2 as totally
126
inert, recyclable and non-toxic solvent for carotenoids extraction, after few decades of evolution, is
127
now used both at the laboratory and industrial scales. The possibility to tune CO2 properties (density,
128
viscosity, diffusivity) by playing on pressure & temperature makes it possible to produce “organic
129
solvent-free” essential oils/oleoresins, mainly from vegetal matrices. But this extraction method is
130
somewhat less selective than extractions based on organic solvents and still requires the use of
131
about 10% w/w of ethanol as cosolvent and further liquid/liquid extractions for carotenoids isolation.
132
Despite its interesting features, in the last 10 years, only few research papers on carotenoids
133
extraction using ASE were published and the first paper dealing with the use of ASE in molecules
134
extraction from food matrices is still quite recent (Breithaupt, 2004). ASE was for instance used for
135
extraction of carotenoids from carrot by-products (Mustafa, Trevino & Turner, 2012), from peppers
136
(Barbero, Palma, Barroso, 2006), from beverages, tomato paste and cereals (Breithaupt, 2004), and
137
from microalgae (Koo, Cha, Song, Chung & Pan, 2012; Cha, Lee, Koo, Song, Lee & Pan, 2010; Denery,
138
Dragull, Tang & Li, 2004). Unfortunately, these articles do not describe accurately the food matrices
139
submitted to extraction processes. For instance, humidity of fresh (non-dehydrated) tissues minced
140
in homogenizers is never given, which can however greatly influence the extraction performances.
141
With regards to dehydrated samples, information is also limited to the fact that pieces of the food
142
matrix are freeze-dried and then grinded in order to obtain a powder whose granulometry is not
143
studied. However, granulometry is of obvious importance on the packing surface area (expressed in
144
m-1 as a result of powder surface area divided by the high pressure cell volume) and therefore on the 6
145
contact between the food matrix and the solvent with direct effect on extraction performances.
146
Granulometry below 0.5 mm is recommended by ASE equipment manufacturers (Thermo Fisher
147
Scientific, 2013).
148
The aim of this investigation is to use ASE for determining the major carotenoids content of three
149
fruits cultivated in Tunisia and known for their richness in carotenoids pigments: kaki also known as
150
kaki/Japanese/Asian Persimmon (Diospyros kaki L.), peach (Prunus persica L.), apricot (Prunus
151
armeniaca L.). After an accurate description of our samples preparation, the influence of extraction
152
temperature, solvent composition and number of extraction cycles on extraction yields of
153
carotenoids will be investigated through an experimental design, focusing on four standard
154
carotenoids: three xanthopylls in all-trans form (lutein, zeaxanthin, β-cryptoxanthin) and one
155
carotene (all-trans-β-carotene). Based on LC-MS analysis, recommendations will finally be provided
156
for the ASE extraction of other carotenoids identified in the three fruits matrices.
157
2. Material and Methods
158
2.1. Chemicals
159
All-trans-β-carotene (Type II synthetic, purity > 95%), all-trans-lutein from marigold & all-trans-β-
160
cryptoxanthin (purity > 97%), all-trans-zeaxanthin (purity > 95%), butylated hydroxytoluene (BHT,
161
purity ≥ 99%) and triethylamine (TEA, purity ≥ 99%) were obtained from Sigma-Aldrich (Shnelldorf,
162
Germany). HPLC organic solvents were of analytical grade: methanol (MeOH) was from Carlo Erba
163
(Val-de-Reuil, France), methyl-tert-butyl-ether (MTBE) was from Fisher Scientific (Loughborough, UK),
164
petroleum ether (PE) was from VWR Prolabo (Fontenay-sous-Bois, France). Ultrapure water was
165
obtained from a purified water system Arium® 611UV from Sartorius (Göttingen, Germany) with a
166
resistivity of 18,2 MΩ*cm. Sodium chloride (NaCl, purity > 99%) and potassium hydroxide (KOH,
167
purity > 99%) were obtained from VWR Prolabo (Fontenay-sous-Bois, France). Florisil 100-200 mesh
168
(Sigma-Aldrich, Shnelldorf, Germany). Liquid Nitrogen was from Air Liquide (Nancy, France).
7
169
2.2. Plant materials
170
The present study focused on three fruits cultivated in Tunisia, Diospyros kaki L. (persimmon var.
171
Triumph from the North region), Prunus persica L. (peach var. Caramel from the North region), and
172
Prunus armeniaca L. (apricot var. Chachi from the Middle region). For each specie, 27 fruits were
173
harvested at full maturity from three selected trees. . 2.3. Samples preparation & characterization
174
Selected fruits were immediately washed with distillated water and placed in hermetical polystyrene
175
boxes in which we added some liquid nitrogen in order to prevent oxidation reactions, before to be
176
stored at -20°C.. Five kg of each whole fruits (flesh and peal) were manually cut into approximately 2
177
g pieces, frozen at -80°C and at once freeze-dried during 48 hours with a pilot CryoTec freeze-dryer
178
(Saint-Gély-du-Fesc, France) with a separated cold trap at -70°C, plates temperature set to -20°C and
179
a working pressure of 15 Pa. Freeze-dried fruits pieces were then placed in a desiccator containing
180
P2O5 in order to prevent water sorption. They were weighted before and after dessication for dry
181
matter % determination. After freeze-drying, fruits pieces were immediately grounded with a
182
Moulinex coffee grinder (Ecully, France). Small amount of liquid nitrogen were added to prevent
183
oxidation during grinding. Powders were stored in a desiccator containing P2O5 in order to prevent
184
water sorption and their residual water content was measured after 24 h at 105°C. Particle size
185
distribution of fruits powders were measured through dynamic light scattering with a Malvern
186
Mastersizer 2000 apparatus (Malvern, UK) in absolute ethanol.
187
2.4. Accelerated Solvent Extraction process
188
All Carotenoid extraction experiments were carried out in a static mode with a Dionex ASE 350
189
extractor (Salt Lake City, UT, USA), using 12 stainless steel cells of 22 mL volume. 60 mL amber vials
190
were used for extracts collection. Common extraction parameters were as follows: (i) 2 g of fruit
191
powder was loaded into the cell and residual cell volume was partially filled with 2 mm glass beads;
192
(ii) the cell was filled with solvent to a pressure of 1500 psi (103 bars); (iii) heat was applied for a
193
initial heat-up time of 5 min; (iv) static extraction with all system valves closed was performed with a 8
194
5 min cycle time; (v) the cell was rinsed with 60% of the cell volume with extraction solvent (in case
195
of 5 static cycles were applied, this 60% of the cell volume was then automatically divided in five
196
times 12% of the cell volume); (vi) the solvent was purged from the cell with N2 gaz for 60 s; and (vii)
197
the system was depressurized. Extraction solvents, methanol (MeOH) and THF, were degassed for 30
198
min by ultrasonic bath Sonoclean ® Labo Moderne (Paris, France) before use. For Kaki powder 1 g of
199
Florisil was added as dispersing agent. Although methanol is a toxic solvent, we decided to use it in
200
combination with THF because (i) it was shown to be highly efficient and more efficient than ethanol
201
for carotenoids extractions when used in combination with THF (Rivera & Canela, 2012) and (ii)
202
because the use of ASE does not require large amounts of organic solvents as compared to
203
extractions at atmospheric pressure. Besides, we decided to use methanol combined with THF in
204
various proportions in order to tune the polarity of our solvent mixture in the perspective of
205
xanthophylls (polar carotenoids) extraction. Last but not least, the purpose of the present study is to
206
provide a convenient method for the identification of the carotenoids content in small samples of
207
kaki, peach and apricot powders. The objective is not to generate a large scale extraction method,
208
which would consume large amounts of methanol. This would be contradictory with the aim to limit
209
the use of toxic organic solvents. It is however obvious that, for preparative extractions of large
210
amounts of carotenoids from fruits, only food compatible organic solvent such as ethanol should be
211
used.
212
2.5. Soxhlet extraction
213
In order to evaluate the efficiency of the extraction of carotenoids using the ASE extraction method,
214
we used a reference extraction method based on a classical soxhlet extraction. This type of
215
extraction based on the use of organic solvents at atmospheric pressure is indeed often used as a
216
reference method leading to a total extraction of carotenoids. Two grams of orange colored fruit
217
powder were mixed with two grams of florisil 100-200 mesh and submitted to a soxhlet extraction
218
with 250 mL of solvent (MeOH:THF, 1:1, v/v) during 6 hours at 66°C (boiling temperature of THF at
9
219
atmospheric pressure) up to stabilization of the extract absorbance and strong discoloration of the
220
fruit sample, which turned clear grey. ASE extraction yields were calculated as a percentage, taking
221
soxhlet extraction results as a reference.
222
2.6. Experimental design
223
For each fruit and carotenoid, a two factors-two levels full factorial design with additional central
224
points was applied with JMP 10 software from SAS Institute Inc. (Cary, NC, USA). Temperature was
225
set to 40°C, 60°C, or 80°C; volume % of MeoH in the MeOH:THF binary solvent was set to 20, 60 or
226
100 %. This design contained 7 experiments (4 corner points, 3 central points). For each of the three
227
fruits, experiments were performed in triplicates leading to 7*3=21 extractions (total of 63
228
extractions for the three fruits). Responses to optimize were the respective concentrations, C, in
229
each carotenoid, expressed in µg/g fruit powder. For all-trans-lutein, all-trans-zeaxanthin, all-trans-β-
230
cryptoxanthin and all-trans-β-carotene, surface responses resulting from a first order linear model
231
with simple interaction, as introduced in equation 1, were plotted together with experimental points
232
using Matlab 7.1 software from MathWorks (Natick, Ma, USA). Significance of residuals normality
233
was tested using a Shapiro-Wilk test with a α risk of 0.05. Equation 1 is:
234
ܶ ∙ ܽ = ܥ+ ܾ ∙ ܯ+ ܿ ∙ ܶ ∙ ܯ+ ܿ݁ݐݏ
235
Where C is the concentration found in the fruit powder sample (µg/g fruit powder); T is the
236
temperature (°C); M is the % methanol in the binary solvent methanol/MTBE; a, b, c and d are
237
coefficients to identify.
(1)
238 239
2.7. Standards preparation and calibration curve
240
All standards stock solutions were prepared and kept under nitrogen atmosphere at -20°C until
241
analysis. All-trans-β-carotene (5 mg) was dissolved in 50 mL of hexane, all-trans-β-crypthoxanthin (1
242
mg) was dissolved in 10 mL of hexane, all-trans-zeaxanthin (1 mg) was dissolved in 10 mL of
10
243
chloroform and all-trans-lutein (1 mg) was dissolved in 10 mL of absolute ethanol. Accurate
244
concentration of the stock solution was measured with a Shimadzu UV 3600 spectrophotometer
245
(Kyoto, Japan) using appropriate extinction coefficient (ε) in L.mol−1.cm−1 and wavelength (nm) for
246
each standards as follows: all-trans-β-carotene-2592 at 450 nm in hexane (John Scott, Finglas, Seale,
247
Hart & de Froidmont-Görtz, 1996); all-trans-zeaxanthin-2480 at 451 nm in hexane (Hart & John Scott,
248
1995), all-trans-lutein-2550 at 445 nm in ethanol and all-trans-β-cryptoxanthin- 2460 at 451 nm in
249
hexane (Heinonen, Ollilainen, Linkola, Varo & Koivistoinen, 1989).
250
subdivided into 1 mL aliquots, transferred in amber vials and stored under nitrogen atmosphere at -
251
20 °C before analysis. Working solutions of each standard were daily prepared from aliquots and
252
absorbance was checked. Dilution were made in HPLC injection solvent (MeOH:MTBE, 1:1, v/v) with
253
addition of 0.1 % BHT (w/v) in order to prevent oxidation. Standards curves were built with seven
254
different concentrations for each carotenoid (1.4, 1.6, 2.0, 2.5, 3.3, 5, 10 µg/mL)
255
2.8. Preparation of carotenoids ASE and soxhlet extracts
256
Pressurized liquid carotenoid extracts and soxhlet extracts were prepared for HPLC analysis according
257
to the method described by Dias & al. (2010) with slight modifications and was conducted under
258
limited light. At first, ASE extract was homogenized with a vortex (Scientific Industries Inc., New York,
259
U.S.A ) at 13,500 rpm during 1 min before being filtered under vacuum through a AP 25 glass fiber
260
filter disc of 2 µm porosity (Millipore, Darmstadt, Germany). Then, 10 mL of the filtrate was added to
261
10 mL of 10 % (w/v) aqueous sodium chloride solution and carotenoids were transferred to a 10 mL
262
petroleum ether phase containing 0.1 % BHT (w/v) using a liquid-liquid extraction. This liquid/liquid
263
extraction was repeated at least 3 times in order to achieve a total transfer of carotenoids toward
264
the petroleum ether phase (extract was then totally discolored). Organic phase was subsequently
265
washed three times with 10 mL of ultrapure water in order to remove residual NaCl traces under
266
neutral pH. Petroleum ether was then evaporated at 30 °C under nitrogen flow using a Turbo Vap®
267
LV (Biotage AB, Uppsala, Sweden). The residue was rapidly dissolved in 5 mL of mobile phase
All stock solutions were
11
268
(MeOH:MTBE, 1:1, v/v) containing 0.1 % BHT (w:v) and filtered through 0.45 µm PVDF syringe filters
269
(Pall Life Sciences, Ann Arbor, PN, USA). Then 2 mL of the filtered solution was mixed with 1 mL of
270
MeOH containing 0.1% BHT (w/v) and saponified with 1 mL of 20 % (w/v) methalonic KOH solution
271
under nitrogen, in the dark, for 1 h, at room temperature. Finally 20 µL of the saponified sample was
272
injected for HPLC analysis.
273
2.9. LC-PDA and LC-PDA-MS analysis.
274
Carotenoids analyses were performed by reverse phase chromatography on a Shimadzu HPLC
275
equipment consisting of a LC-20 AD pump, CBM-20A controller, CTO-20A oven and SPD-M20A PDA
276
detector at 450 nm (Shimadzu, France). We used a 2504.6 mm ID S-5 µm YMC C30 columm
277
coupled with a 104.0 mm ID S-5µm guard cartridge columm (ImChem, Versailles, France). Analysis
278
was performed at 25°C. Mobile phase consisted in a gradient of MeOH (A), and MTBE (B). Both
279
solvent A and B contained 0.1% BHT (w/v). . Flow rate was 1.0 mL/min. The gradient profile of the
280
mobile phase was set as follows: linear increasing gradient from 5% B to 30 % B in 30 min; linear
281
increasing gradient from 30 % B to 50 % B in 20 min; decreasing from 50 % B to 5 % B in 0.01 min.
282
The column was equilibrated for 10 min at the starting conditions before each injection. Before each
283
batch of HPLC analysis, stabilization time of the columm was 30 min, and a blank (MeOH: MTBE, 1:1,
284
v/v) was injected. A mixture of our four standards (all-trans-lutein, all-trans-zeaxanthin, all-trans-β-
285
cryptoxanthin, all-trans-β-carotene) was injected, every 4 samples, in order to control eventual
286
retention times deviations.
287
In order to identify additional carotenoids and compare the fruits carotenoids profiles, LC-PDA-MS
288
runs were performed in positive APCI mode with a Shimadzu LCMS-2020 mass spectrometer
289
(Shimadzu, France). A scanning rate of 2143 u/s was used in the range 50-2000 amu. Nebulizing gaz
290
flow was fixed to 1.5 mL/min. Interface voltage and temperature were 4.5 kV and 250°C,
291
respectively. The elution profile of the mobile phase was the same as with HPLC alone. The column
292
was equilibrated for 5 min at the starting conditions before each injection. 12
293
2.10. Limits of detection and quantification
294
The detection and quantification (LOD and LOQ) were determined for all standards according to a
295
US-EPA method (Hubaux & Vos, 1970). LOD and LOQ were calculated as [6*sa0 /a] and [10*sa0 /a]
296
respectively, with sa0 and a the standard deviation on the intercept and the slope of the calibration
297
curve, respectively.
298
3. Results and discussion
299
3.1. Sample characterization prior to ASE
300
Particle size distribution (see additional figure not included in the manuscript) of the three fruits
301
powders showed that these powders are almost totally (> 95% of the total volume) made of particles
302
below 500 µm as recommended for ASE extractions on powder. Particles size ranged from 1 µm to
303
about 1000 µm. Based on triplicates, measured moisture content of fruits (skin + flesh) , expressed as
304
g of water per 100 g of fresh weight, were found to be 77.66 ± 1.63 %, 85.85 ± 2.79 % and 87.00 ± 5.08
305
% for kaki (Diospyros kaki), peach (Prunus persica) and apricot (Prunus armeniaca), respectively. The
306
significantly lower water content of kaki as compared to peach and apricot is in accordance with data
307
from the Danish food composition databank (www.foodcomp.dk), which indicates that this lower
308
water content is balanced by a higher carbohydrate content of about 23 g/g of fresh product in kaki
309
against only about 10 g/g of fresh product in peach and apricot. This could explain why we
310
experienced clogging when running preliminary ASE experiments with kaki powder, especially at
311
80°C. In order to prevent this clogging (sticky aspect at the bottom of the extraction cell) we
312
therefore mixed our 2 g of kaki powder with 1 g of florisil (100-200 mesh).
313
3.2. Calibration curves and quantifications
314
R² values of 0.99 were achieved with the four quantified carotenoids. Limit of quantification
315
(detection) were found to be 0.41 (0.25), 0.21 (0.13), 0.25 (0.15) and 0.63 (0.38) µg/mL for all-trans-
13
316
lutein at 11.9 min, all-trans-zeaxanthin at 14.2 min, all-trans-β-cryptoxanthin at 22.9 min, and all-
317
trans-β-carotene at 33.9 min, respectively.
318 319
3.3. Carotenoids extraction in kaki, peach and apricot
320
Table 1 lists the surfaces responses model coefficients and their associated p-value for each fruit-
321
carotenoid combination. Figure 1 presents the experimental points and surfaces responses
322
representing the amount, in µg, of all-trans-lutein (1a, 1b, 1c) and all-trans-zeaxanthin (1d, 1e, 1f)
323
extracted per g of fruit powder. Figure 2 presents the experimental points and surfaces responses
324
representing the amount, in µg, of all-trans-β-cryptoxanthin (2a, 2b, 2c) and all-trans-β-carotene (2d,
325
2e, 2f) extracted per g of fruit powder. Figure 3 presents carotenoids profiles obtained by HPLC with
326
kaki, peach and apricot, after extraction under the optimal conditions identified in our experimental
327
domain. In any cases, the variability of carotenoids extractions as a function of temperature and % of
328
methanol in the solvent could be appropriately explained by a first order linear model with
329
interaction between temperature and methanol % (R²-values ranging from 0.77 up to 0.92). p-values
330
much lower than 0.05 showed that temperature, % methanol but also the interaction between these
331
two factors clearly influenced the extraction yield whatever the fruits and carotenoids under
332
concern.
333
Several observations can be made from the surface responses. At first, when using a high extraction
334
temperature (80°C), the binary solvent composition did not influence the carotenoids extraction,
335
whatever the fruit. Similarly when using 100% methanol as a solvent, excepted with all-trans-β-
336
carotene in apricot, temperature did not or slightly affected the extraction of carotenoids. In addition
337
and surprisingly, both xanthophylls, whether they are dioxygenated (all-trans-lutein and all-trans-
338
zeaxanthin) or monoxygenated (all-trans-β-crypotoxanthin), and all-trans-β-carotene, were all better
339
extracted using MeOH:THF, 20:80, v/v, at 40°C. With regard to the solvent, we expected (i) a better
340
extraction of xanthophylls with higher proportions (60 or 100%) of the protic and polar solvent that is
14
341
methanol; (ii) a better extraction of all-trans-β-carotene when using higher proportions of the aprotic
342
and moderately polar solvent that is THF. Nevertheless it was not the case as presented in figures 1
343
and 2, which clearly show that, at 40°C and when using high pressure extraction (103 bars), THF is a
344
more efficient solvent than methanol for xanthophylls as well as for a carotene such as all-trans-β-
345
carotene.
346
Although carotenoids content of a fruit can be affected by many factors such as variety, agronomic
347
technics and climatic conditions, post-harvest conditions and resulting maturity, our results are quite
348
in accordance with recent published studies where liquid extractions at atmospheric pressures were
349
used in place of ASE extraction (see Table 2). For instance, based on a recent analysis of carotenoids
350
content in 32 different astringent kaki cultivars where carotenoids content was expressed in µg / 100
351
g of fresh weight, and considering a % dry matter of 23 %, it was found that all-trans-β-cryptoxanthin,
352
followed by all-trans-zeaxanthin, were the most abundant carotenoids in kaki, which we also
353
observed in our astringent Diospyros kaki L. var. Triumph (Zhou, Zhao, Sheng, Tao & Yang, 2011). In
354
peach we observed that all-trans-β-carotene, followed by all-trans-β-cryptoxanthin, were the most
355
abundant carotenoids, which is in accordance with recent data from Campbell & Padilla-Zakour on
356
three different peach varieties (2013). In apricot, as currently and recently observed (Fratianni,
357
Albenese, Mignogna, Cinquenta, Panfili & Di Matteo, 2013), all-trans-β-carotene is by far the most
358
abundant carotenoid with more than 70 % of the sum of all-trans-lutein, all-trans-zeaxanthin, all-
359
trans-β-cryptoxanthin and all-trans-β-carotene. In addition and according to table 2 it seems that
360
ASEis an appropriate method in order to achieve a better extraction of all-trans-lutein, as compared
361
to extractions at atmospheric pressures observed by other authors. But this higher concentration of
362
lutein that we observed can also be due to other steps such as the way we performed our transfer of
363
extracted lutein in petroleum ether, the way we performed our saponification step which is crucial to
364
hydrolyze lutein esters and therefore detect lutein, and finally the HPLC conditions. We used a
365
250*4.6 mm C30 column with a long gradient program, which is much more resolutive as compared
366
to a C18 column with similar dimensions. With a C18 column, when one attempts to separate both 15
367
polar (xanthophylls) and non polar (carotenes) carotenoids in one single HPLC run, it is difficult to
368
find appropriate gradient and highly polar xanyhophylls such as lutein may be not retained by the
369
column, being directly lost in the injection peak without possible quantification. In our case, as
370
presented below, we could for instance also detect violaxanthin, which is more polar than lutein.3.4.
371
Influence of extraction cycles number on extraction yield
372
In order to assess the influence of the number of static extraction cycles on carotenoids extraction
373
yields, we performed extractions based on 1 or 5 static extraction cycles under the best conditions
374
identified in our experimental domain (Methanol : THF, 20:80, v/v and 40 °C). Table 3 shows average
375
and standard deviations of the extraction yields (n=3) and mention whether the difference between
376
these average yields is significant or not (p