Chemical Engineering and Processing - Process Intensification 163 (2021) 108359
5
was used as a positive control and the antioxidant index used to evaluate
the antioxidant activity is AI 50 , defined as extract concentration needed
to consume 50 %, of the electro generated radical corresponding to (Ipa 0
- Ipa S )/Ipa 0 = 0.5. To estimate AI 50 the variation of (1- Ipas/Ipa 0 ) were
plotted according to the concentration of added saponins (Fig. 3).
For the measurements of the decreasing of oxidation peak current of
O 2
• −
in presence of extract we adopted the method used by Guendouze
et al. [21] which aims at exploiting the convolution time semi-derivative
transformation of cyclic voltammograms. Indeed, for convolution
curves, the baselines are easier to define automatically in comparison
with the asymmetric cyclic voltammetric curves and the oxidation peak
current values, before and after extract additions, can be more accurately measured (Fig. S.1).
2.8. Identification of saponins by ESI-HR-MS measurements
In order to confirm the molecular weight of saponins, extract by MAE
from E. sepositus, high resolution mass spectrometry experiments (HRESI-MS) were performed on a Thermo Fisher Scientific Q-Exactive mass
spectrometer, in both positive and negative ionization modes by direct
infusion. MS experiments were performed at the higher resolution of the
instrument (140,000 at m/z 200). Elementary compositions of the ions
were determined using instrument software (Xcalibur).
The MAE extract was prepared using a concentration of 10–20 μg
mL
− 1
and the flow rate was 5− 10 μL min
− 1
. ESI parameters were as
follows: spray voltage 3− 4 kV, capillary temperature 300
◦
C, sheath gas
flow rate 10-80, auxiliary gas flow 0-10, SLens RF Level 100.
The single-stage ESI–MS spectra were recorded by scanning
150− 2000 m/z mass range.
2.9. Statistical analysis
In our work, all experiments were carried out in triplicate and the
results have been reported as means ± SD.
The analysis of the influence of the single-factor experiment, of the
data obtained from the BBD trials and the influence of extraction technique (MAE, UAE or CSE) on total saponins compound (TSC) yield, was
statistically assessed by ANOVA and Tukey’s posthoc test for means
discrimination (95 % confidence level).
To test the model significance for the MAE the data obtained from
BBD for the response variable were statistically analyzed using ANOVA.
The model was taken significant and highly significant at p < 0.05 and p
< 0.01, respectively.
The Minitab software was used to construct the BBD experimental
design and to analyze all the results.
3. Results and discussion
3.1. Microwave-assisted extraction
3.1.1. Single-factor experiments
The saponins recovery and the experimental conditions of the
various TSC values are summarised in Table 1. It can be observed, that
the higher values of TSC correspond to the aqueous methanol (24 mg
g
− 1
).
Table 1 shows, that TSC yield increase when methanol percentage
increases from 20 to 50 % at constant microwave power, extraction time
and liquid-solid ratio. These results mean that at higher methanol concentrations, absorbed microwave energy increases, consequently, the
solvent power to dissolve saponins also increases.
Eskilsson et al. (2000) reported that the differences of TSC yield are
attributed to the difference in dielectric properties of the solvent which
significantly influence the ability of a solvent to absorb microwave energy [1].
Methanol has a high dissipation factor (tan δ = 6400 10
− 4
), and a
relatively high dielectric constant (ε’ = 33.62) [1]. Hence, the aqueous
methanol can be considered as appropriate solvent for saponins
extraction from starfish by MAE.
A methanol 50 % (v/v) was set for the next single-factor experiments
and the range of 30–70 % (v/v) was selected for the optimization of MAE
by the RSM.
Concerning the effect of irradiation, the TSC increased significantly
from 39 to 55 mg g
− 1
between 1–3 min, and decreases at longer irradiation times (4–8 min) with a highly significant difference (Table 1). In
fact, longer irradiation by microwave induces probably thermal degradation of bioactive compounds [15,22]. The range of 1–5 min has been
selected for the optimization of saponins extraction by MAE.
In case of microwave power, the TSC yield increased significantly as
microwave power has increased from 100 to 200 W, then decreased with
rising microwave power in tested conditions (Table 1). This effect of
microwave power on TSC yield could be due to its heating effect, which
induces increase of mass transfer phenomena, up to a certain microwave
power value, and then thermal degradation of bioactive compounds at
higher microwave power [15,22]. According to the obtained results, the
200 W power have been used for the last single-factor trials, while the
range 100− 300 W was selected for the RSM study.
Table 1 shows that the ratio liquid-solid significantly affected the
TSC values, in way that as increasing liquid-solid ratio from 10 to 40 mL
g
− 1
, the TSC yield raised significantly, however higher volumes failed to
increase TSC yield. This result can be explained by the fact that, the
solvent volume must be just sufficient to immerse the entire sample [1].
The adequate volume of solvent increases the concentration gradient to
drive the mass transfer. Based on statistical analysis, the range of 20 to
50 mL g
-1
was selected for the RSM optimization.
3.1.2. Response surface methodology optimization of operating parameters
The response surface methodology (RSM) optimization has been
directed to the TSC compounds involved in this present work. The
experimental values of the TSC obtained from BBD trials and the corresponding predicted values according to the applied second-order
regression model are shown in Table 2. The statistical analysis
(ANOVA) appropriate to the second order model (Eq. (1)) is reported in
Table 3.
The ANOVA analysis demonstrated that the interaction term
extraction time and ratio (X 2 X 4 ) was not significant (p > 0.399), the
other linear terms, quadratic and interaction was highly significant (p <
0.01), hence the mathematical model (Eq. (4)) correlating the recovery
of TSC with MAE process variables is given below excluding nonsignificant terms: mg g
− 1
.
Ysap
(
mg g
− 1
)
= 191.66 + 3.5527X 1 + 26.56X 2 + 0.32X 3 + 3.65X 4
− 0.0597X 1 X 2 − 0.0001X 1 X 3 + 0.0089X 1 X 4 − 0.0112X 2 X 3
− 0.0022X 3 X 4 − 0.03287X
2
1 − 3.5182X
2
2 − 0.0005X
2
3
− 0.0433X
2
4
(4)
To determine whether the accuracy of mathematical model, the ANOVA
analysis was conducted. The results of the statistical analysis of the
second order model fitting are illustrated in Table 3. The ANOVA
demonstrated that the model was highly significant.
The determination coefficient (R
2
) value of 0.9998 indicates that
only 0.02 % of the total variations was not explained by the model. In
addition, the adjusted determination coefficient (R
2
adj = 0.9997) is too
close to R
2
. Moreover, the value of R
2
predicted is 0.9996, it is in
agreement with the value of R
2
adj. All of these figures confirm that the
model is highly significant.
The lack of fit test indicates that the fitting model is adequate to
describe the experimental data with the value of lack of fit test of 0.215
higher than 0.05 and not significant relative to the pure error.
Eq. (4) was used to generate the three-dimensional response surface
plots (Fig. 1). The plots were obtained by plotting the in the Z-axis the
B. Dahmoune et al.
5
was used as a positive control and the antioxidant index used to evaluate
the antioxidant activity is AI 50 , defined as extract concentration needed
to consume 50 %, of the electro generated radical corresponding to (Ipa 0
- Ipa S )/Ipa 0 = 0.5. To estimate AI 50 the variation of (1- Ipas/Ipa 0 ) were
plotted according to the concentration of added saponins (Fig. 3).
For the measurements of the decreasing of oxidation peak current of
O 2
• −
in presence of extract we adopted the method used by Guendouze
et al. [21] which aims at exploiting the convolution time semi-derivative
transformation of cyclic voltammograms. Indeed, for convolution
curves, the baselines are easier to define automatically in comparison
with the asymmetric cyclic voltammetric curves and the oxidation peak
current values, before and after extract additions, can be more accurately measured (Fig. S.1).
2.8. Identification of saponins by ESI-HR-MS measurements
In order to confirm the molecular weight of saponins, extract by MAE
from E. sepositus, high resolution mass spectrometry experiments (HRESI-MS) were performed on a Thermo Fisher Scientific Q-Exactive mass
spectrometer, in both positive and negative ionization modes by direct
infusion. MS experiments were performed at the higher resolution of the
instrument (140,000 at m/z 200). Elementary compositions of the ions
were determined using instrument software (Xcalibur).
The MAE extract was prepared using a concentration of 10–20 μg
mL
− 1
and the flow rate was 5− 10 μL min
− 1
. ESI parameters were as
follows: spray voltage 3− 4 kV, capillary temperature 300
◦
C, sheath gas
flow rate 10-80, auxiliary gas flow 0-10, SLens RF Level 100.
The single-stage ESI–MS spectra were recorded by scanning
150− 2000 m/z mass range.
2.9. Statistical analysis
In our work, all experiments were carried out in triplicate and the
results have been reported as means ± SD.
The analysis of the influence of the single-factor experiment, of the
data obtained from the BBD trials and the influence of extraction technique (MAE, UAE or CSE) on total saponins compound (TSC) yield, was
statistically assessed by ANOVA and Tukey’s posthoc test for means
discrimination (95 % confidence level).
To test the model significance for the MAE the data obtained from
BBD for the response variable were statistically analyzed using ANOVA.
The model was taken significant and highly significant at p < 0.05 and p
< 0.01, respectively.
The Minitab software was used to construct the BBD experimental
design and to analyze all the results.
3. Results and discussion
3.1. Microwave-assisted extraction
3.1.1. Single-factor experiments
The saponins recovery and the experimental conditions of the
various TSC values are summarised in Table 1. It can be observed, that
the higher values of TSC correspond to the aqueous methanol (24 mg
g
− 1
).
Table 1 shows, that TSC yield increase when methanol percentage
increases from 20 to 50 % at constant microwave power, extraction time
and liquid-solid ratio. These results mean that at higher methanol concentrations, absorbed microwave energy increases, consequently, the
solvent power to dissolve saponins also increases.
Eskilsson et al. (2000) reported that the differences of TSC yield are
attributed to the difference in dielectric properties of the solvent which
significantly influence the ability of a solvent to absorb microwave energy [1].
Methanol has a high dissipation factor (tan δ = 6400 10
− 4
), and a
relatively high dielectric constant (ε’ = 33.62) [1]. Hence, the aqueous
methanol can be considered as appropriate solvent for saponins
extraction from starfish by MAE.
A methanol 50 % (v/v) was set for the next single-factor experiments
and the range of 30–70 % (v/v) was selected for the optimization of MAE
by the RSM.
Concerning the effect of irradiation, the TSC increased significantly
from 39 to 55 mg g
− 1
between 1–3 min, and decreases at longer irradiation times (4–8 min) with a highly significant difference (Table 1). In
fact, longer irradiation by microwave induces probably thermal degradation of bioactive compounds [15,22]. The range of 1–5 min has been
selected for the optimization of saponins extraction by MAE.
In case of microwave power, the TSC yield increased significantly as
microwave power has increased from 100 to 200 W, then decreased with
rising microwave power in tested conditions (Table 1). This effect of
microwave power on TSC yield could be due to its heating effect, which
induces increase of mass transfer phenomena, up to a certain microwave
power value, and then thermal degradation of bioactive compounds at
higher microwave power [15,22]. According to the obtained results, the
200 W power have been used for the last single-factor trials, while the
range 100− 300 W was selected for the RSM study.
Table 1 shows that the ratio liquid-solid significantly affected the
TSC values, in way that as increasing liquid-solid ratio from 10 to 40 mL
g
− 1
, the TSC yield raised significantly, however higher volumes failed to
increase TSC yield. This result can be explained by the fact that, the
solvent volume must be just sufficient to immerse the entire sample [1].
The adequate volume of solvent increases the concentration gradient to
drive the mass transfer. Based on statistical analysis, the range of 20 to
50 mL g
-1
was selected for the RSM optimization.
3.1.2. Response surface methodology optimization of operating parameters
The response surface methodology (RSM) optimization has been
directed to the TSC compounds involved in this present work. The
experimental values of the TSC obtained from BBD trials and the corresponding predicted values according to the applied second-order
regression model are shown in Table 2. The statistical analysis
(ANOVA) appropriate to the second order model (Eq. (1)) is reported in
Table 3.
The ANOVA analysis demonstrated that the interaction term
extraction time and ratio (X 2 X 4 ) was not significant (p > 0.399), the
other linear terms, quadratic and interaction was highly significant (p <
0.01), hence the mathematical model (Eq. (4)) correlating the recovery
of TSC with MAE process variables is given below excluding nonsignificant terms: mg g
− 1
.
Ysap
(
mg g
− 1
)
= 191.66 + 3.5527X 1 + 26.56X 2 + 0.32X 3 + 3.65X 4
− 0.0597X 1 X 2 − 0.0001X 1 X 3 + 0.0089X 1 X 4 − 0.0112X 2 X 3
− 0.0022X 3 X 4 − 0.03287X
2
1 − 3.5182X
2
2 − 0.0005X
2
3
− 0.0433X
2
4
(4)
To determine whether the accuracy of mathematical model, the ANOVA
analysis was conducted. The results of the statistical analysis of the
second order model fitting are illustrated in Table 3. The ANOVA
demonstrated that the model was highly significant.
The determination coefficient (R
2
) value of 0.9998 indicates that
only 0.02 % of the total variations was not explained by the model. In
addition, the adjusted determination coefficient (R
2
adj = 0.9997) is too
close to R
2
. Moreover, the value of R
2
predicted is 0.9996, it is in
agreement with the value of R
2
adj. All of these figures confirm that the
model is highly significant.
The lack of fit test indicates that the fitting model is adequate to
describe the experimental data with the value of lack of fit test of 0.215
higher than 0.05 and not significant relative to the pure error.
Eq. (4) was used to generate the three-dimensional response surface
plots (Fig. 1). The plots were obtained by plotting the in the Z-axis the
B. Dahmoune et al.