Fourier Transform Infrared (FT-IR) Spectroscopy as a Possible Rapid Tool to Evaluate Abiotic Stress E ﬀ ects on Pineapple By-Products

: Fourier transform infrared (FT-IR) spectroscopy is a physicochemical technique based on the vibrations of a molecule energized by infrared radiation at a speciﬁc wavelength range. Abiotic stresses can induce the production of secondary metabolites, increasing bioactivity. The objectives of the study were to evaluate the impact of heat treatments on the bioactivity of pineapple by-products, and whether FT-IR analysis allows understanding of the changes imparted by abiotic stress. The by-products were treated at 30, 40, and 50 ◦ C for 15 min, followed by storage at 5 ± 1 ◦ C for 8 and 24 h. Lyophilized samples were characterized for total phenolic content and antioxidant capacity and analyzed by FT-IR. Thermal treatments at 50 ◦ C reduced the content of phenolic compounds (21–24%) and antioxidant capacity (20–55%). Longer storage time (24 h) was advantageous for the shell samples, although this e ﬀ ect was not demonstrated for the core samples. The principal components analysis (PCA) model developed with the spectra of the pineapple shell samples showed that the samples were grouped according to their total phenolic compounds content. These results allow the conclusion to be drawn that FT-IR spectroscopy is a promising alternative to the conventional chemical analytical methodologies for phenolic and antioxidant contents if there are signiﬁcant di ﬀ erences among samples.


Introduction
Infrared spectroscopy is one of the most important analytical techniques in existence, able to analyze practically any sample in almost any state. Fourier transform infrared (FT-IR) spectroscopy is an instrumental method based on measurement of the vibration of a molecule excited by infrared radiation at a specific wavenumber range. Infrared spectroscopy has advantages over some conventional techniques. This technique is non-destructive, fast and simple to use, precise, mechanically simple, and can be used for routine quantitative and qualitative analysis. The spectroscopic techniques are inexpensive and they do not require time-consuming sample pre-treatment or the use of (environmentally harmful) chemical extracts [1,2]. In the near infrared and mid-infrared regions, an element normally absorbs at more than one wavelength, and the absorbance at a presented wavelength
The control samples were maintained in the same conditions, but the heat treatment was not applied: Ctr_0, Ctr_8, and Ctr_24.
After storage time, all samples were frozen at −80 °C and lyophilized (Telstar Lyo Quest, Telstar, Spain). Treatments were performed in duplicate in two independent composite samples.
Thermal treatments: The bags with wounded by-products were submerged in a water bath (Selecta, Spain) at temperatures of 30, 40, and 50 • C for 15 min and stored at 5 ± 1 • C for 8 and 24 h: T30_8,  T30_24, T40_8, T40_24, T50_8, T50_24. The control samples were maintained in the same conditions, but the heat treatment was not applied: Ctr_0, Ctr_8, and Ctr_24.
After storage time, all samples were frozen at −80 • C and lyophilized (Telstar Lyo Quest, Telstar, Spain). Treatments were performed in duplicate in two independent composite samples.

Total Phenolic Content and Antioxidant Capacity
The extract preparation involved making a ratio of 1:10 (m:v) of sample and methanol (Sigma-Aldrich, Germany), followed by Ultra-Turrax homogenizer (IKA LABORTECHNIK T25 basic, Janke & Kunkel GmbH &Co., Germany) at 8000 rpm for 2 min and incubation overnight at 4 • C. The extracts were obtained by centrifugation (HERMLE Z383K LABORTECHNIK, Germany) at 8000 rpm for 20 min (4 • C), and the supernatants were stored at 4 • C protected from light until needed for analysis.
The total phenolic content (TPC) was determined according to the method of Swain and Hills (1959) [24] and Heredia and Cisneros-Zevallos (2009) [25], with some alterations. Aliquots of the extract supernatant after centrifugation (150 µL) were diluted with 2400 µL nanopure water, followed by the addition of 150 µL of 0.25 M Folin-Ciocalteu (Panreac AppliChem, Germany), and the sample was incubated for 3 min at room temperature. The reaction was interrupted by adding 300 µL of 1M Na 2 CO 3 (Panreac AppliChem, Germany) and the mixture was incubated for a further 2 h, protected from light. Afterwards, the absorbance of the solution was measured at 725 nm. The total phenolic content was defined using a standard curve developed with equivalent gallic acid (GAE) and expressed as mg GAE.g −1 dry weight. The average of three replicates was used for each condition.
The antioxidant capacity was evaluated by DPPH (2,2-diphenyl-1-picrylhydrazyl) method following the procedure of Brand-Williams et al. (1995) [26], with some modifications. In this case, the DPPH solution was formulated with methanol until it achieved 1.1 units of absorbance at 515 nm. The sample extracts were prepared as described above. Sample aliquots of 100 µL were taken from the supernatants and then added to 3900 µL DPPH solution. This mixture was homogenized, and the reaction took place for 40 min in the dark. The sample absorption was read at 515 nm. The blank was prepared with methanol and used as control and to calibrate the spectrophotometer for readings. The antioxidant capacity was determined using a standard curve developed with Trolox, and the results are expressed as Trolox equivalent antioxidant capacity (TEAC; µmol Trolox.g −1 dry weight). The average of three replicate samples was used for each condition.
Ferric-reducing antioxidant power (FRAP) was performed according to Benzie and Strain (1996) [27], with some modifications. Solutions of 300 mM acetate buffer (3.1 g sodium acetate (C 2 H 3 NaO 2 ·3H 2 O; Panreac AppliChem, Germany) and 16 mL acetic acid glacial (C 2 H 4 O 2 ; FisherChemical (United Kingdom)), pH 3.6, 10 mM TPTZ (2,4,6-tripyridyl-s-triazine; Sigma-Aldrich, Germany) in 40 mM hydrochloric acid (HCl; Sigma-Aldrich, Germany), and 20 mM iron (III) chloride (FeCl 3 ; Sigma-Aldrich, Germany) were prepared. The working solution was made by combining 35 mL acetate buffer 300 mM, 3.5 mL TPTZ solution, and 3.5 mL FeCl 3 solution. The procedure involved mixing 2.7 mL of the FRAP solution with 270 µL H 2 O and 90 µL of the extract samples, which had been warmed in a water bath at 37 • C for 30 min. The coloured result (ferrous tripyridyltriazine complex) was then read at 595 nm using water as the blank. The antioxidant capacity was determined using a standard curve established with Trolox, and the results are expressed as Trolox equivalent antioxidant capacity (TEAC; µmol Trolox.g −1 dry weight). The average of three replicates was used for each condition.
The antioxidant capacity was also measured using the ABTS (2,2 -azino-bis(3-ethylbenzothiazoline-6-sulphonic acid)) method as described by Re et al. (1999) and Rufino et al. (2007) [28,29], with some modifications. Two stock solutions were prepared: ABTS solution (7 mM) and potassium persulfate solution (140 mM). The working solution was prepared by mixing 2 mL of ABTS solution with 35.2 µL of the potassium persulfate solution and keeping it in the dark at room temperature for 12-16 h. The ABTS solution was then diluted with methanol to reach an absorbance of 0.700 at 734 nm. The reaction was performed by mixing 2970 µL ABTS solution with 30 µL sample aliquots for 6 min and the absorbance at 734 nm was immeditately recorded. The absorbance of the reaction samples was related to the Trolox standard and the results are expressed in terms of Trolox equivalent antioxidant capacity (TEAC; µmol Trolox.g −1 dry weight). The average of three replicate samples was used for each condition.

Fourier Transform Infrared Spectroscopy
A FT-IR spectrometer Spectrum Two (Perkin-Elmer, USA) with a diamond ATR (attenuated total reflection) single reflection accessory was used. PerkinElmer Spectrum software was used to draw the spectra. The spectra (32 scans per spectrum) of the lyophilized pineapple by-products were collected in duplicate in the mid-infrared wavenumber range from 4000 to 400 cm −1 , with a spectral resolution of 4 cm −1 .

Statistical Analysis
The results obtained in the analytical assays were subjected to statistical analysis using Statistica TM v.8 Software (StatSoft Inc., USA). Statistically significant differences (p < 0.05) between samples were defined using Tukey's honestly significant difference test.
The results were also submitted to principal component analysis (PCA) using Statistica TM v.8 Software. In this case, all variables were mean-centered and standardized (scaled) to unit variance prior to analysis [30]. Matlab version 7.9 (MathWorks, Natick, MA, USA) and the PLS Toolbox version 4.0 (Eigenvector Research Inc.-USA) for Matlab were utilized to perform the qualitative analysis of the spectra using PCA, according to the description presented elsewhere [3,31]. To derive the PCA models, only the region of 600-4000 cm −1 was used, in order to exclude noise and the non-informative range of the spectra. To enhance the correlations concerning radiation absorption and the properties of each sample, different preprocessing methods were evaluated. The choice was made by analyzing the ones that allowed the highest data variance with the lowest number of PCs to be captured with the smaller errors, using cross-validation and leave-one-out method as the internal validation strategy. In this work, baseline correction followed by mean-centering were always used as spectrum pre-processing methods.

In Pineapple Shell and Core
As shown in Table 1, pineapple shell samples (65.75 mg GAE.g −1 dry weight) presented significantly higher levels (p < 0.05) of phenolic compounds than core (59.89 mg GAE.g −1 dry weight). Antioxidant capacity, determined by the three tested methods (DPPH, FRAP, and ABTS) was also higher in the shell samples. Table 1. Total phenolic content (TPC) and antioxidant capacity of the wounded shell and core pineapple samples submitted to thermal treatments evaluated by three different methods (DPPH-2,2-diphenyl-1-picrylhydrazyl; FRAP-ferric-reducing antioxidant power; ABTS-2,2azino-bis(3-ethylbenzothiazoline-6-sulphonic acid). As the main objective of the current work was to evaluate the use of mid-infrared spectroscopy to study the impact of abiotic stresses on the amount of bioactive compounds, wounded samples without thermal treatment, control samples, were analyzed for total phenolic compounds and antioxidant capacity using conventional methods. As shown in Table 1, the control samples after 8h (Ctr_8) and 24 h (Ctr_24) of storage time presented significantly higher (p < 0.05) levels of phenolic compounds than the initial control samples (Ctr_0). This indicates that storage time after cutting improves the synthesis of phenolic compounds. The same behavior was observed for antioxidant capacity, determined by the three tested methods (DPPH, FRAP, and ABTS). The enhancement of bioactive compounds in fresh and treated fruits by applying abiotic stresses (wounding, heat shock, UV irradiation, among others) has been reported previously by Heredia and Cisneros-Zevallos (2009). The wound-induced increase of phenolic compounds has been suggested to be due to phenylpropanoid pathway activation [13].

Thermal Treatment Effect in Pineapple Shell and Core
As shown in Table 1, thermal treatment for 15 min affected the total phenol content and antioxidant capacity of the samples. The samples heat-treated at 30 or 40 • C for 15 min (T30_15 and T40_15) showed significantly higher (p < 0.05) phenolic content and antioxidant capacity than control samples. The thermal treatment at 40 • C during 15 min (T40_15) seemed to be the treatment that promoted the highest synthesis of phenolic compounds. Results are not consistent as to which storage time after thermal treatment, 8 or 24 h, was more effective. The storage time studied was short, to verify more evident differences.
On the other hand, the thermal treatments at 50 • C (T50) showed lower phenolic compounds and antioxidant capacity for all storage times tested. This results indicates that the temperature was excessive and decreased phenylalanine-ammonia lyase (PAL) activity, which is in accordance with other authors [32].

Fourier Transform Infrared Spectroscopy
Responding to the main objective of the research, to evaluate the use of mid-infrared spectroscopy to study the impact of abiotic stresses, samples were also analyzed using FT-IR.
The spectra of Ctr_8 and Ctr_24 are shown in Figure 2. The band presented at 3600-3000 cm −1 , with a maximum value close to 3300 cm −1 was associated with the stretching vibration of O-H groups (3600-3200 cm −1 ), at 3400-3300 cm −1 with N-H stretching, and at 3100-3000 cm −1 with =C-H stretching [33,34]. The peak at 1719 cm −1 was associated with carbonyl group C=O stretching. Since the samples were lyophilized, the O-H stretching band should have been mainly associated with compounds such as carboxylic acids, alcohols, and phenols [34]. In fact, according to Table 1, the TPC values of the shell control samples, which had more intense bands in this region, were higher (6.4%). On the other hand, the band 3000-2800 cm −1 , with a maximum near 2930 cm −1 , corresponded to the C-H stretching of CH 2 groups. This band is usually associated with carbohydrates and fats [35]. The bands in the 1800-1500 cm −1 region corresponded to C=C and C=O stretching. In the pineapple sample, these bonds may have been related to proteins, amide I (1700-1600 cm −1 ) and amide II (1565-1520 cm −1 ), and fats (1745-1725 cm −1 ) [34].
(6.4%). On the other hand, the band 3000-2800 cm −1 , with a maximum near 2930 cm −1 , corresponded to the C-H stretching of CH2 groups. This band is usually associated with carbohydrates and fats [35]. The bands in the 1800-1500 cm −1 region corresponded to C=C and C=O stretching. In the pineapple sample, these bonds may have been related to proteins, amide I (1700-1600 cm −1 ) and amide II (1565-1520 cm −1 ), and fats (1745-1725 cm −1 ) [34]. Several marker bands were identified in the spectral range of 1400-800 cm −1 . These bands may have been associated with the stretching and bending of carbohydrates. The bands in the region of 1150-900 cm −1 were assigned to C-O and C-C stretch modes, while those in the 1400-1200 cm −1 region were due to O-C-H, C-C-H, and C-O-H bending vibrational modes of the carbohydrates [36]. In addition to sugars, bands characteristic of organic acids (malic, citric, and ascorbic acids) might also have been observed in the absorption region 1500-900 cm −1 [37]. The bands between 900-600 cm −1 corresponded to a fingerprint region with vibrational C-H or CH2 group deformation. The absorption region at 750-500 cm −1 may have been associated with =C-H bending or C-C bending vibrations that occur at low frequencies (below 500 cm −1 ). The absorption region at 700-720 cm −1 was related to colour pigments (β-carotene) [34,38].
The score plots of the principal component analysis of the PCA model established with all the FT-IR spectra of the shell and core samples are shown in Figure 3 and Figure 4, respectively. In Figure 3b, the loadings plot of the first principal component (PC1) of the shell samples is also Several marker bands were identified in the spectral range of 1400-800 cm −1 . These bands may have been associated with the stretching and bending of carbohydrates. The bands in the region of 1150-900 cm −1 were assigned to C-O and C-C stretch modes, while those in the 1400-1200 cm −1 region were due to O-C-H, C-C-H, and C-O-H bending vibrational modes of the carbohydrates [36]. In addition to sugars, bands characteristic of organic acids (malic, citric, and ascorbic acids) might also have been observed in the absorption region 1500-900 cm −1 [37]. The bands between 900-600 cm −1 corresponded to a fingerprint region with vibrational C-H or CH 2 group deformation. The absorption region at 750-500 cm −1 may have been associated with =C-H bending or C-C bending vibrations that occur at low frequencies (below 500 cm −1 ). The absorption region at 700-720 cm −1 was related to colour pigments (β-carotene) [34,38].
The score plots of the principal component analysis of the PCA model established with all the FT-IR spectra of the shell and core samples are shown in Figures 3 and 4, respectively. In Figure 3b, the loadings plot of the first principal component (PC1) of the shell samples is also presented. As shown, this plot presented the most significant peaks at around 3300 cm −1 , which related to the combination bands of O-H bonds, and at 1000 cm −1 , which were assigned to the O-C-H, C-C-H, and C-O-H vibrational modes of the carbohydrates. This result confirmed that PC1 captured the change of the TPC of the samples. The loadings plot with respect to core samples was similar to the one obtained for the shell samples, with two important peaks at around 3300 cm −1 and 1000 cm −1 (data not shown).
As shown in Figures 3 and 4, PC1 captured 99.08% and 98.75% of the spectra variance for shell and core samples, respectively. In shell samples (Figure 3a), it was observed that the model separated the samples with higher contents of phenolic compounds (Ctr_8, Ctr_24, T30_24, and T40_8), from the samples with lower contents (T30_8, T40_24, T50_8, and T50_24). Figure 3 also indicates that the FT-IR allowed detection of the differences between the samples in terms of the TPC and antioxidant capacity, and for results consistent with the conventional analyses to be obtained. The loadings plot of the model PC1, presented in Figure 3b, permitted the identification of the wavenumbers that were the most important to describe PC1. It is worth noting that FT-IR has also been applied to evaluate the quality of kiwi samples, and the results showed relatively high correlation between the values of the antioxidant capacity measured by the conventional methods and FT-IR spectroscopy [20]. related to the combination bands of O-H bonds, and at 1000 cm −1 , which were assigned to the O-C-H, C-C-H, and C-O-H vibrational modes of the carbohydrates. This result confirmed that PC1 captured the change of the TPC of the samples. The loadings plot with respect to core samples was similar to the one obtained for the shell samples, with two important peaks at around 3300 cm −1 and 1000 cm −1 (data not shown).     O-C-H, C-C-H, and C-O-H vibrational modes of the carbohydrates. This result confirmed that PC1 captured the change of the TPC of the samples. The loadings plot with respect to core samples was similar to the one obtained for the shell samples, with two important peaks at around 3300 cm −1 and 1000 cm −1 (data not shown).      On the other hand, as shown in Figure 4, most of the samples presented similar scores on PC1, between −0.5 and 0, whereas only the control sample with 8 h (8) of storage time was separated from the others. This result is a clear indication that the small differences in the phenolic compounds content and antioxidant capacity (Table 1) were not enough to lead to significant changes in the FT-IR spectra. Therefore, in the scores plot of the PCA model illustrated in Figure 4, the samples had similar scores on PC1 and, contrary to what was discussed above for the shell samples, it was not possible to find any distribution pattern.
However, the utilization of FT-IR spectroscopy to establish the polyphenols in durian, mango, and avocado samples has been described in the literature. The authors stated that these analytical methods might be applicable for phytochemical analysis in other samples [22].

Conclusions
The present study allowed the conclusion to be drawn that abiotic stresses, wounding, and thermal treatments applied to pineapple by-products (shell and core) samples induced the synthesis of phenolic compounds.
The use of FT-IR analysis based on the content of phenolic compounds and antioxidant capacity was only possible if there are significant differences among samples. It was possible for materials like the pineapple shell samples, which presented high phenol contents and reacted enough to the abiotic stress applied.