Leaves of C. uvifera were collected at Tecolutla, Veracruz, Mexico and transferred to the Laboratorio Integral de Investigación en Alimentos (LIIA) of the Tecnológico Nacional de México (Tepic, Nayarit, México). They were washed and dried until reaching 5% humidity. Subsequently, they were ground and sieved (1 mm, pore size). A homogeneous green powder was obtained and stored in a desiccator at 25℃, until the extraction process.

N-hexane (Jalmek, Nuevo León, Mexico); ethyl acetate; resazurin sodium salt were purchased from MP Biomedicals, LLC. (Illkirch, France). 96° ethanol was obtained from Guinama (Spain). Caco-2 cell lines; Eagle’s Minimum Essential Medium (EMEM); fetal bovine serum; trypsin-EDTA; were purchased from ATCC^® (Manassas, VA, USA). Penicillin-streptomycin, amphotericin-B, lipopolysaccharides (LPS), lupeol standard, PEO (Mw of 400,000) and silica gel (60 Å, 40–63 µm) were obtained all from Sigma Aldrich^® (St. Louis, MO, USA). The surfactant TEGO SML was obtained from Evonik Industries AG. (Germany). HDPAF were obtained in the LIIA of ITT. Deionized water was used throughout the study.

The extraction conditions were optimized in previous works conducted by Ramos-Hernández et al. and Cruz-Salas et al. [8, 9]. To obtain the extract, the methodology reported by Cruz-Salas et al. (2022) [9] was used. Briefly, the powdered plant material was mixed with n-hexane in a 1:10 ratio (w/v), and the mixture was sonicated at 42 kHz at 25℃ for 30 min in a CD-4820 Kendal Digital Ultrasonic Cleaner (China) and vacuum filtered using Whatman^™ 1 paper filters (110 mm, diameter). The solvent was removed on a rotary evaporator IKA RV10 (Wilmington, NC, USA) at 200 rpm at 45℃. The extract obtained was placed in an extraction hood to eliminate solvent residues and stored in a desiccator until further use.

The purification process for obtaining a lupeol-rich fraction was performed by column chromatography. A vertical chromatographic column 50 cm long and 1.5 cm of diameter was used. Silica gel was used as stationary phase and gradients of 100 mL hexane (100%), 200 mL hexane-ethyl acetate (95:5), 100 mL hexane-ethyl acetate (80:20) were used. The stationary phase was conditioned with 100 mL of hexane (100%), and then, the C. uvifera extract previously diluted (1:20 ratio) in hexane was added. The gradients were successively fed to the column, which facilitated the transfer of the compounds through the stationary phase, and an average volume of 1 mL of effluent per sample was continuously collected.

A qualitative identification of the different fractions was carried out using the TLC technique, which has been previously used to identify fractions rich in lupeol from the hexane extract of Solanum melongena L [12]. Silica gel plates on TLC Al foils (20 cm × 20 cm, 60 Å medium pore diameter, Sigma-Aldrich^®, Steinheim, Germany) were used as stationary phase, using the mixture of hexane, ethyl acetate and dichloromethane in ratio 80:10:10 respectively, as mobile phase. The identification of the fractions was carried out by placing on a plate the references of the standards of the compounds to be identified, the extract and the samples obtained from the elution. Samples with same signals on TLC were considered as the same fraction. Solvent was removed from the obtained fractions, and the resultant dried fractions were stored in sterile vials in a desiccator until further analysis.

A Bruker Tensor 37 FT-IR spectrometer (Bruker, Ettlingen, Germany) coupled with the ATR sampling accessory Golden Gate^™ (Specac Ltd., Orpington, UK) was used to compare the ATR-FTIR spectra of the lupeol-rich fraction and lupeol standard. All spectra were recorded between 4,000 and 600 cm^–1, with a resolution of 4 cm^–1, and averaging 10 scans. The software Opus 4.0 (Bruker, Ettlingen, Germany) was used to analyze the spectral data. Same methodology was used for the analysis of the electrospun fibers encapsulating the lupeol-rich fraction.

NMR analysis was performed using the high-resolution one-dimensional technique on a Bruker AVANCE III 400 MHz equipment (Bruker Corporation am Silberstreifen, Rheinstetten, Germany). 10 mg of the sample were dissolved in 0.5 mL of deuterated chloroform (CDCL₃) the mixture was filtered before being analyzed. Samples were analyzed in triplicates. Spectra were analyzed with Bruker TopSpin 3.2 software (Bruker Corporation am Silberstreifen, Rheinstetten, Germany).

A Waters^® ACQUITY^® TQD equipment (Milford, M.A, USA) with the multiple reaction monitoring (MRM) mode was used. 5 µL of the fraction was injected into the liquid chromatography-mass spectrometry (LC-MS) using the autosampler. A Luna^® Omega polar^® C18 UHPLC column, (1.6 µm; 100 × 2.1 mm) (Phenomenex) was used at 40℃. A flow of 0.4 mL/min was applied. The mobile phase consisted of a gradient of acidified water (0.1% formic acid): acetonitrile. The analysis was carried out by applying the gradient at different times: 35:65 (0 min), 95:05 (1.2 min), 35:65 (3.3 min), respectively. The lupeol standard was used to develop the calibration curve (y = 1.8558x + 22.0701, r² = 0.9925, where x is the concentration in ng/mL and y is the area under the curve) in the concentration range from 10 to 1,000 ng/mL with LOD: 0.8421 ng/mL and LOQ: 2.5518 ng/mL. Spectra were obtained with positive and negative ionization, MRM mode and the transition of lupeol was 427.0 > 315.2, using Systems supported on MassLynx 4.1 or Empower 2 Software (Milford, MA, USA). Samples were analyzed in triplicates.

Measurements were performed by dissolving 10 mg of extract in 1 mL of methanol, 50 µL of the stock solution were taken and diluted with 950 µL of methanol. 0.1 mL of solution were mixed with 1.9 mL of 2,2-diphenyl-1-picrylhydrazyl (DPPH) 0.094 mM in methanol and stored protected from the light for 30 min. The absorbance was read at 517 nm in a Cary 50-Bio UV-Vis spectrophotometer (Varian, Australia PTY). The DPPH radical inhibition was calculated using equation 1:

DPPH inhibition (%) = (A – B) / A × 100 (equation 1)

Where A is the DPPH absorbance and B is the sample absorbance. Samples were analyzed in triplicates.

Different concentrations of the lupeol-rich fraction at 10 µg/mL, 20 µg/mL, 50 µg/mL and 100 µg/mL were evaluated. As a positive control of viability, medium without the lupeol-rich fraction was used. Once the incubation time had elapsed, 50 µL of resazurin solution were added, and cells were incubated for 3 h at 37℃ and 5% CO₂. Resorufin formation by cellular metabolic activity was determined in a CLARIOstar^® Plus plate reader (BMG LABTECH, Germany) determining fluorescence (excitation at 560 nm and emission at 590 nm). The results were obtained with the MARS^® version 3.30 software (BMG LABTECH, Germany). The percentage of cellular cytotoxicity was determined (equation 2), considering the untreated cells as a positive control (0% cytotoxicity):

Cytotoxicity (%) = (F_trat – F_b) / (F_cont – F_b) × 100 (equation 2)

Where F_trat is the treatment fluorescence, F_b is the blank fluorescence, and F_cont is the control fluorescence. Samples were analyzed in triplicates.

The nanofibers morphology was characterized by scanning electron microscopy (SEM) using a SEM Hitachi S-4800 (Hitachi High Technologies Corporation, Tokyo, Japan). Samples were sputtered with a gold-palladium coating for 2 min, and SEM micrographs were taken at 10 kV. The size distribution of the structures was determined measuring at least 100 measurements per sample.

The human colon adenocarcinoma cell line (Caco-2) was used to evaluate the lupeol-rich fraction permeability. These cells were cultured as explained in previous section. After 21 days, the monolayer integrity was checked by measuring the trans-epithelial electrical resistance [TEER (Ω·cm²)] with a Millicell-ERS-2 volt-ohm-meter (Merck-Millipore, Darmstadt, Germany). The apparent permeability of loaded lupeol-rich fraction loaded in electrospun nanofibers was evaluated at a concentration of 20 µg/mL of the lupeol-rich fraction and compared with the pure fraction at the same concentration as a control. Samples were analyzed in triplicates.

Apparent permeability (P_app) coefficient (expressed in cm/s) was calculated according to equation 3:

P _app = [V / (A × T)] (d_r / d_d) (equation 3)

Where V is the volume (mL) of the receptor compartment, A is the area (cm²) of the Transwell membrane, T is the time (s), d_r is the concentration of lupeol in the receptor compartment, and d_d is the concentration of lupeol in the donor compartment, both expressed in (ng/mL).

Results showed in this work are the average and standard deviation of at least three replicates. The data were analyzed by applying a one-way analysis of variance (ANOVA) followed by a Tukey mean comparison test (P ≤ 0.05). STATISTICA software version 10 (StatSoft, Inc., Tulsa, OK, USA) was used.

The C. uvifera extract was fractionated by means of a chromatographic column in a total of 9 fractions. The composition of the different fractions eluted from the column was preliminary studied by TLC based on the comparison with the pure compounds, i.e., lupeol and α-amirin. In Figure 1, it shows the TLC signals of lupeol and α-amirin standards and their comparison with fractions 5 and 6. Based on the similarity in signals and color intensity of the lupeol standard with fraction 6 as shown in Figure 1, it is possible to preliminarily ascertain that fraction 6 showed the highest abundance of lupeol.

After fractionation, fraction 6 was subjected to solvent evaporation to concentrate the solid extract. During this process, the formation of small needles was observed.

Identification of lupeol in fraction 6 was made by comparison of the ATR-FTIR spectrum of fraction 6 with the spectrum of lupeol standard. The ATR-FTIR spectrum of the lupeol standard, shown in Figure 2A, presented characteristic bands at ca. 3,280 cm^–1 and 1,043 cm^–1 ascribed to O-H bond vibrations, ca. 2,931 and 2,872 cm^–1 ascribed to the C-H bond in alkanes, ca. 1,640 cm^–1 ascribed to the stretching vibration of the C=C bond, ca. 1,450 cm^–1 ascribed to the bending vibration of the C-H bond in the cyclohexyl group, ca. 894 cm^–1 related to the bending vibrations of the C=C-H bond.

According to ATR-FTIR spectrum of fraction 6 shown in Figure 2B, it showed the main characteristic peaks of pure lupeol, confirming the abundance of this molecule in this fraction.

NMR analysis was performed to elucidate the structural bonds for the identification of lupeol in fraction 6. In Figure 3A, it shows the NMR-1H one-dimensional spectrum of the lupeol compound (C₃₀H₅₀O). The hydrogen deficiency index (HDI) was determined obtaining a value of 6; which indicates the possible presence of double bonds. Proton signals were obtained with displacement at δ 0.768, 0.808, 0.924, 0.948, 1.009, and 1.367 ppm, a sextuplet with signals at δ 1.862, 1.879, 1.893, 1.902, 1.916, 1.933 ppm, which could correspond to the cyclic alkanes present in the structure of lupeol. On the other hand, the signals corresponding to the carbonyl group were identified with a quintuplet at δ 2.338, 2.357, 2.367, 2.375, 2.385 ppm. The doublet at δ 4.671 ppm could correspond to the shift of olefinic hydrogen bound to C-29. A singlet (δ 4.9 ppm) was also observed, which could correspond to the H of the OH group of the lupeol molecule located at C-3 (Figure 3A). In relation to the analysis of fraction 6 (Figure 3B), the presence of lupeol was confirmed by studying the one-dimensional NMR of 1H spectrum. The displacements observed in the spectrum of the fraction showed analogous signals to those obtained in the spectrum of the lupeol standard. Displacement signals were observed at δ 0.759, 0.786, 0.826, 0.942, 0.966, 1.027, 1.251, 1.679 ppm, the cyclic alkanes at δ 1.879, 1.897, 1.919, 1.934, 1.951 ppm, the presence of the carbonyl group at δ 2.384 ppm, considering a correlation between both spectra.

The LC-MS analysis allowed the identification and quantification of lupeol in fraction 6. Based on its reference standard, lupeol presented a retention time of 2.79 min. As shown in Figure 4, same retention time was observed also for fraction 6, confirming the presence of lupeol in this fraction. In addition, mass spectrum of fraction 6 is in agreement with the mass spectrum of lupeol standard. According to the calibration curve, the abundance of lupeol in fraction 6 of the C. uvifera leaf extract was 0.77 mg per gram of dried fraction 6 (77%).

The analysis of the free radical scavenging capacity DPPH revealed that the hexane extract of C. uvifera leaves presented a 24.6% ± 0.6% inhibition of the DPPH radical. This result is in agreement with the result reported for the fruit pulp of C. uvifera with the same method, i.e. 22.8% of DPPH inhibition [13]. However, the inhibition of the DPPH radical decreased to a 7.3% ± 0.1% for the fraction 6. Since fraction 6 has a higher concentration of lupeol than the extract, its antioxidant activity is lower.

This assay was performed to determine the non-toxic concentrations range of fraction 6 to perform later on the permeability test. According to the results shown in Figure 5, concentrations of fraction 6 up to 50 µg/mL did not provoke a decrease on cell viability. It is important to note that values slightly over 100% obtained at 10 µg/mL and 20 µg/mL were probably due to cell growth during the assay. However, 10% decrease in the viability of the cells was observed at 50 µg/mL, despite this decrease was not statistically significant. In contrast, a significant decrease of a 39% in the percentage of cell viability was observed when applying 100 µg/mL of fraction 6 (P ≤ 0.05).

These results were also corroborated by analyzing the cell morphology by optical microscopy (Figure 6). Important changes in cells morphology were observed in Figure 6a to Figure 6d, which were directly proportional to the increase of the concentration of fraction 6, and especially significant when applying the highest concentration of 100 µg/mL of fraction 6. The Figure 6a shows the typical appearance of a Caco-2 cell monolayer, confirming that the concentration used of fraction 6 did not have a significant effect on the cells. However, in Figure 6b and Figure 6c, stress signs were detected. Some cells started losing their adherence to the substrate due to the breakage of adhesion proteins present on the membranes. In Figure 6d, signs of apoptosis were detected including fragmentation and condensation of the cells with preservation of their organelles, to finally observe the breaking up of the cells into membrane-bound apoptotic bodies. However, this apoptosis hypothesis should be proved in further studies.

Electrospun nanofibers were obtained through the electrospinning process as shown in Figure 7. Neat polymeric mixtures produced homogeneous and continuous nanofibers with a rough surface as shown in Figure 7a, with an average fiber diameter of 248 nm ± 46 nm. When incorporating the fraction into the solution, no significant differences were observed in fiber diameter, since the average size for the nanofibers encapsulating the fraction was 254 nm ± 57 nm.

In Figure 8, it shows the comparison of the spectra of the HDPAF/PEO electrospun nanofibers with and without fraction 6. Band assignation for the lupeol-rich fraction was already analyzed in section “Identification of lupeol by ATR-FTIR”.

In the spectrum of the HDPAF/PEO nanofibers, intense signals were observed, the vibration at 2,281 cm^–1 corresponding to an asymmetric stretching of the C-H bonds of CH₂, the vibration at 1,093 cm^–1 associated to the stretching vibration of the C-O bonds, both signals are related to the presence of PEO, as it has been reported by Abdelrazek et al. [14] for pure PEO. The HDPAF were identified with the vibration at 1,670–1,645 cm^–1 associated to the particular β-hydroxy aldehyde stretching of the C=O bond characteristic of the structure of agave fructans signals and an intense broadband between 3,000–3,500 cm^–1, related to the C-H bond stretching, and O-H bond stretching characteristic of the hydroxyl groups. Similar characteristic bands were reported by Ramos-Hernández et al. (2018) [8] for agave fructans processed by electrospraying.

The nanofibers with the encapsulated fraction showed the characteristic bands of HDPAF and PEO. Regarding the characteristic bands of the fraction, it was possible to detect the band at 1,680–1,620 cm^–1 assigned to C=C bonds, characteristic of the lupeol molecule. The intense band of the fraction around 2,900 cm^–1, which is associated with the presence of C-H bonds could not be detected in the fraction-loaded nanofibers, since this band remained hindered below the signal of the neat polymers.

The Caco-2 cells permeability of the pure fraction of the C. uvifera extract was compared to the permeability of the fraction of the C. uvifera extract encapsulated in HDPAF/PEO electrospun nanofibers. This test was performed at 20 µg of fraction/mL. Experimental apparent permeability coefficients (P_app, cm/s) were calculated for the time-dependent in vitro absorption from the apical side of the Caco-2 monolayer to the basal side of the Caco-2 monolayer.

The results of Table 1 indicate that, 15 min after the start of the assay, the electrospun nanofibers loaded with fraction 6 presented increased lupeol permeability in comparison to the pure fraction, which did not present a response at this time. This increase in permeability may be related to the smaller drug size inside the nanofibers.